-
Notifications
You must be signed in to change notification settings - Fork 0
Expand file tree
/
Copy pathindex.html
More file actions
6 lines (6 loc) · 116 KB
/
Copy pathindex.html
File metadata and controls
6 lines (6 loc) · 116 KB
1
2
3
4
5
6
<!DOCTYPE html><html lang="en"><head><meta charSet="utf-8"/><meta name="viewport" content="width=device-width, initial-scale=1"/><link rel="preload" as="image" href="/internvla-a15.github.io/reference-assets/paper/teaser.png"/><link rel="stylesheet" href="/internvla-a15.github.io/_next/static/chunks/025_jc0.9ca-..css" data-precedence="next"/><link rel="preload" as="script" fetchPriority="low" href="/internvla-a15.github.io/_next/static/chunks/0emxhq679m-p2.js"/><script src="/internvla-a15.github.io/_next/static/chunks/0v2xy4nvj4cxc.js" async=""></script><script src="/internvla-a15.github.io/_next/static/chunks/12mz0_._155bq.js" async=""></script><script src="/internvla-a15.github.io/_next/static/chunks/turbopack-02-_ioaox7weq.js" async=""></script><script src="/internvla-a15.github.io/_next/static/chunks/072b-ukyyvq~h.js" async=""></script><script src="/internvla-a15.github.io/_next/static/chunks/05.ce42pp9k.4.js" async=""></script><script src="/internvla-a15.github.io/_next/static/chunks/03xdlv9wrei1r.js" async=""></script><meta name="theme-color" content="#fbfbfd" media="(prefers-color-scheme: light)"/><meta name="theme-color" content="#0d0f17" media="(prefers-color-scheme: dark)"/><meta name="color-scheme" content="light dark"/><title>InternVLA-A1.5</title><meta name="description" content="InternVLA-A1.5 unifies vision-language understanding, latent visual foresight, and action generation for robot policy learning."/><meta name="application-name" content="InternVLA-A1.5"/><meta name="creator" content="InternVLA-A1.5 Team"/><meta property="og:title" content="InternVLA-A1.5"/><meta property="og:description" content="InternVLA-A1.5 unifies vision-language understanding, latent visual foresight, and action generation for robot policy learning."/><meta property="og:site_name" content="InternVLA-A1.5"/><meta property="og:type" content="website"/><meta name="twitter:card" content="summary_large_image"/><meta name="twitter:title" content="InternVLA-A1.5"/><meta name="twitter:description" content="InternVLA-A1.5 unifies vision-language understanding, latent visual foresight, and action generation for robot policy learning."/><script src="/internvla-a15.github.io/_next/static/chunks/03~yq9q893hmn.js" noModule=""></script></head><body><div hidden=""><!--$--><!--/$--></div><script>((a,b,c,d,e,f,g,h)=>{let i=document.documentElement,j=["light","dark"];function k(b){var c;(Array.isArray(a)?a:[a]).forEach(a=>{let c="class"===a,d=c&&f?e.map(a=>f[a]||a):e;c?(i.classList.remove(...d),i.classList.add(f&&f[b]?f[b]:b)):i.setAttribute(a,b)}),c=b,h&&j.includes(c)&&(i.style.colorScheme=c)}if(d)k(d);else try{let a=localStorage.getItem(b)||c,d=g&&"system"===a?window.matchMedia("(prefers-color-scheme: dark)").matches?"dark":"light":a;k(d)}catch(a){}})("class","theme","light",null,["light","dark"],null,false,true)</script><header class="sticky top-0 z-40 border-b border-border/70 bg-background/82 backdrop-blur-xl"><div class="mx-auto grid h-18 max-w-[1120px] grid-cols-[1fr_auto_1fr] items-center px-5"><a class="flex items-center gap-3 justify-self-start font-semibold" href="#top"><span aria-hidden="true" class="size-9 rounded-xl bg-[linear-gradient(135deg,#005ea8,#0d74ce_52%,#2aa8f2)] shadow-sm"></span><span class="text-lg">InternVLA-A1.5</span></a><nav class="hidden items-center gap-7 justify-self-center text-sm font-medium text-muted-foreground lg:flex"><a class="transition hover:text-foreground" href="#introduction">Introduction</a><a class="transition hover:text-foreground" href="#model">Model</a><a class="transition hover:text-foreground" href="#data">Data</a><a class="transition hover:text-foreground" href="#benchmarks">Benchmarks</a><a class="transition hover:text-foreground" href="#experiments">Experiments</a><a class="transition hover:text-foreground" href="#bibtex">BibTeX</a></nav><div class="justify-self-end"><button class="inline-flex shrink-0 items-center justify-center gap-2 whitespace-nowrap rounded-full border text-sm font-medium transition-all outline-none focus-visible:ring-3 focus-visible:ring-ring/30 disabled:pointer-events-none disabled:opacity-50 [&_svg]:pointer-events-none [&_svg]:size-4 [&_svg]:shrink-0 active:translate-y-px border-border/80 bg-card/70 text-foreground backdrop-blur-md hover:bg-muted size-9 p-0" type="button" aria-label="Toggle theme"><svg xmlns="http://www.w3.org/2000/svg" width="24" height="24" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" class="lucide lucide-moon dark:hidden" aria-hidden="true"><path d="M20.985 12.486a9 9 0 1 1-9.473-9.472c.405-.022.617.46.402.803a6 6 0 0 0 8.268 8.268c.344-.215.825-.004.803.401"></path></svg><svg xmlns="http://www.w3.org/2000/svg" width="24" height="24" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" class="lucide lucide-sun hidden dark:block" aria-hidden="true"><circle cx="12" cy="12" r="4"></circle><path d="M12 2v2"></path><path d="M12 20v2"></path><path d="m4.93 4.93 1.41 1.41"></path><path d="m17.66 17.66 1.41 1.41"></path><path d="M2 12h2"></path><path d="M20 12h2"></path><path d="m6.34 17.66-1.41 1.41"></path><path d="m19.07 4.93-1.41 1.41"></path></svg></button></div></div></header><main><section class="hero-section px-5 pb-12 pt-14 sm:pt-18" id="top"><div aria-hidden="true" class="hero-background"></div><div class="mx-auto max-w-[1120px] text-center"><h1 class="hero-title text-balance"><span class="title-gradient">InternVLA-A1.5</span></h1><p class="hero-subtitle">Unifying Understanding, Latent Foresight and<br/>Action for Compositional Generalization</p><div class="authors-line"><span class="author-entry"><strong class="author-name block">Physical Intelligence Team, Shanghai AI Laboratory</strong></span></div><div aria-label="Affiliations" class="team-logo-row"><img alt="Shanghai Artificial Intelligence Laboratory" loading="lazy" width="672" height="390" decoding="async" data-nimg="1" class="team-logo-image" style="color:transparent" src="/internvla-a15.github.io/reference-assets/images/ailab-logo.png"/><img alt="InternRobotics" loading="lazy" width="739" height="110" decoding="async" data-nimg="1" class="team-logo-image" style="color:transparent" src="/internvla-a15.github.io/reference-assets/images/InternRobotics-logo.png"/></div><div class="mx-auto mt-6 flex max-w-4xl flex-wrap justify-center gap-3"><a class="inline-flex shrink-0 items-center justify-center gap-2 whitespace-nowrap rounded-full border text-sm font-medium transition-all outline-none focus-visible:ring-3 focus-visible:ring-ring/30 disabled:pointer-events-none disabled:opacity-50 [&_svg]:pointer-events-none [&_svg]:size-4 [&_svg]:shrink-0 active:translate-y-px border-transparent bg-primary text-primary-foreground shadow-sm hover:bg-primary/90 h-10 px-5" href="/internvla-a15.github.io/reference-assets/paper/InternVLA_A1_5.pdf"><svg xmlns="http://www.w3.org/2000/svg" width="24" height="24" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" class="lucide lucide-file-text" aria-hidden="true"><path d="M6 22a2 2 0 0 1-2-2V4a2 2 0 0 1 2-2h8a2.4 2.4 0 0 1 1.704.706l3.588 3.588A2.4 2.4 0 0 1 20 8v12a2 2 0 0 1-2 2z"></path><path d="M14 2v5a1 1 0 0 0 1 1h5"></path><path d="M10 9H8"></path><path d="M16 13H8"></path><path d="M16 17H8"></path></svg>Paper</a><a class="inline-flex shrink-0 items-center justify-center gap-2 whitespace-nowrap rounded-full border text-sm font-medium transition-all outline-none focus-visible:ring-3 focus-visible:ring-ring/30 disabled:pointer-events-none disabled:opacity-50 [&_svg]:pointer-events-none [&_svg]:size-4 [&_svg]:shrink-0 active:translate-y-px border-border/80 bg-card/70 text-foreground backdrop-blur-md hover:bg-muted h-10 px-5" href="https://github.com/InternRobotics/InternVLA-A-series"><img alt="" aria-hidden="true" loading="lazy" width="16" height="16" decoding="async" data-nimg="1" class="size-4 shrink-0 object-contain invert dark:invert-0" style="color:transparent" src="/internvla-a15.github.io/reference-assets/images/github-invertocat-white.svg"/>Github</a><a class="inline-flex shrink-0 items-center justify-center gap-2 whitespace-nowrap rounded-full border text-sm font-medium transition-all outline-none focus-visible:ring-3 focus-visible:ring-ring/30 disabled:pointer-events-none disabled:opacity-50 [&_svg]:pointer-events-none [&_svg]:size-4 [&_svg]:shrink-0 active:translate-y-px border-border/80 bg-card/70 text-foreground backdrop-blur-md hover:bg-muted h-10 px-5" href="https://huggingface.co/collections/InternRobotics/internvla-a15"><img alt="" aria-hidden="true" loading="lazy" width="16" height="16" decoding="async" data-nimg="1" class="size-4 shrink-0 object-contain" style="color:transparent" src="/internvla-a15.github.io/reference-assets/images/huggingface-logo.svg"/>Huggingface</a><a class="inline-flex shrink-0 items-center justify-center gap-2 whitespace-nowrap rounded-full border text-sm font-medium transition-all outline-none focus-visible:ring-3 focus-visible:ring-ring/30 disabled:pointer-events-none disabled:opacity-50 [&_svg]:pointer-events-none [&_svg]:size-4 [&_svg]:shrink-0 active:translate-y-px border-border/80 bg-card/70 text-foreground backdrop-blur-md hover:bg-muted h-10 px-5" href="https://www.modelscope.cn/collections/InternRobotics/InternVLA-A15"><img alt="" aria-hidden="true" loading="lazy" width="16" height="16" decoding="async" data-nimg="1" class="size-4 shrink-0 object-contain" style="color:transparent" src="/internvla-a15.github.io/reference-assets/images/modelscope-logo.svg"/>ModelScope</a><a class="inline-flex shrink-0 items-center justify-center gap-2 whitespace-nowrap rounded-full border text-sm font-medium transition-all outline-none focus-visible:ring-3 focus-visible:ring-ring/30 disabled:pointer-events-none disabled:opacity-50 [&_svg]:pointer-events-none [&_svg]:size-4 [&_svg]:shrink-0 active:translate-y-px border-border/80 bg-card/70 text-foreground backdrop-blur-md hover:bg-muted h-10 px-5" href="#bibtex"><svg xmlns="http://www.w3.org/2000/svg" width="24" height="24" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" class="lucide lucide-file-text" aria-hidden="true"><path d="M6 22a2 2 0 0 1-2-2V4a2 2 0 0 1 2-2h8a2.4 2.4 0 0 1 1.704.706l3.588 3.588A2.4 2.4 0 0 1 20 8v12a2 2 0 0 1-2 2z"></path><path d="M14 2v5a1 1 0 0 0 1 1h5"></path><path d="M10 9H8"></path><path d="M16 13H8"></path><path d="M16 17H8"></path></svg>BibTeX</a></div><figure class="hero-teaser"><img alt="Overview of InternVLA-A1.5, combining vision-language understanding, latent visual foresight, and action generation." width="2320" height="1493" decoding="async" data-nimg="1" class="h-auto w-full rounded-md bg-white" style="color:transparent" src="/internvla-a15.github.io/reference-assets/paper/teaser.png"/><figcaption>InternVLA-A1.5 uses learnable foresight tokens to query future-relevant latent information from shared multimodal context, while action queries decode continuous control through a lightweight expert.</figcaption></figure><div class="stat-band"><div class="stat-card"><p>98.9</p><h2>LIBERO average</h2><span>Best average success rate on the four LIBERO suites.</span></div><div class="stat-card"><p>84.8</p><h2>LIBERO-Plus total</h2><span>Best total robustness under visual and layout perturbations.</span></div><div class="stat-card"><p>93.2</p><h2>RoboTwin average</h2><span>Strong bimanual manipulation performance across clean and randomized splits.</span></div><div class="stat-card"><p>861M</p><h2>Training frames</h2><span>Robot manipulation corpus spanning 1.2M episodes.</span></div></div></div></section><section class="border-t border-border px-5 py-18 sm:py-24" id="introduction"><div class="mx-auto max-w-[1120px]"><div class="mb-10 max-w-4xl"><p class="section-eyebrow mb-3">Introduction</p><h2 class="text-balance text-3xl font-semibold sm:text-5xl">A Quick Overview of InternVLA-A1.5</h2><p class="mt-5 text-lg leading-8 text-muted-foreground">Start with the project video before the architecture walkthrough.</p></div><figure class="overflow-hidden rounded-lg border border-border bg-card p-4 shadow-sm sm:p-6"><video class="intro-video" controls="" height="1080" playsInline="" poster="/internvla-a15.github.io/videos/intro-poster.jpg" preload="metadata" width="1920"><source src="/internvla-a15.github.io/videos/intro.mp4" type="video/mp4"/>Your browser does not support this video.<!-- --> <a href="/internvla-a15.github.io/videos/intro.mp4">Download the introduction video.</a></video></figure></div></section><section class="border-t border-border px-5 py-18 sm:py-24 bg-muted/35" id="model"><div class="mx-auto max-w-[1120px]"><div class="mb-10 max-w-4xl"><p class="section-eyebrow mb-3">Model</p><h2 class="text-balance text-3xl font-semibold sm:text-5xl">Mixture-of-Transformers with training-only latent foresight</h2><p class="mt-5 text-lg leading-8 text-muted-foreground">The model adopts a Mixture-of-Transformers design with a native VLM backbone and a smaller unified expert that receives multimodal context through shared full attention.</p></div><figure class="overflow-hidden rounded-lg border border-border bg-card p-4 shadow-sm sm:p-6"><div class="tf-root" data-mode="training"><div class="tf-toolbar"><div aria-label="Diagram mode" class="tf-toggle" role="group"><button aria-pressed="true" type="button">Training</button><button aria-pressed="false" type="button">Inference</button></div><div class="tf-controls"><button aria-label="Pause animation" class="tf-icon-btn" type="button"><svg xmlns="http://www.w3.org/2000/svg" width="24" height="24" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" class="lucide lucide-pause size-4" aria-hidden="true"><rect x="14" y="3" width="5" height="18" rx="1"></rect><rect x="5" y="3" width="5" height="18" rx="1"></rect></svg></button></div></div><div class="tf-scroll"><svg aria-label="Token flow through the InternVLA-A1.5 Mixture-of-Transformers: multimodal tokens enter the VLM, pass through six repeated interleaved patterns of one Full Attention layer and three Gated DeltaNet layers, cross into the unified expert through shared full attention, and are decoded into a continuous action chunk via flow matching." class="tf-svg" data-flow-mode="training" data-playing="false" role="img" viewBox="0 0 1000 494" xmlns="http://www.w3.org/2000/svg"><defs><linearGradient gradientUnits="userSpaceOnUse" id="tf-flow-gradient" x1="890" x2="890" y1="113" y2="455"><stop class="tf-flow-gradient-clean" offset="0%"></stop><stop class="tf-flow-gradient-noise" offset="100%"></stop></linearGradient></defs><rect class="tf-tower tf-tower--vlm is-active" height="222" rx="14" width="390" x="50" y="170"></rect><rect class="tf-tower tf-tower--expert" height="222" rx="14" width="360" x="545" y="170"></rect><text class="tf-text tf-text--title" text-anchor="middle" x="245" y="204">Pre-trained VLM</text><text class="tf-text--sub" text-anchor="middle" x="245" y="224">Qwen-3.5 · 2B</text><text class="tf-text tf-text--title" text-anchor="middle" x="725" y="204">Unified Expert</text><text class="tf-text--sub" text-anchor="middle" x="725" y="224">460M</text><path class="tf-link is-active" d="M 92 438 V 392" id="tfg-in-images"></path><path class="tf-link is-active" d="M 200 438 V 392" id="tfg-in-instruction"></path><path class="tf-link is-active" d="M 300 438 V 392" id="tfg-in-mode"></path><path class="tf-link is-active" d="M 382 438 V 392" id="tfg-in-state"></path><path class="tf-link tf-link--dashed" d="M 384 279 H 604" id="tfg-shared-lr"></path><path class="tf-link" d="M 160 170 V 130" id="tfg-out-qa"></path><path class="tf-link tf-training-only" d="M 395 170 V 130" id="tfg-out-fast"></path><path class="tf-link tf-link--rose" d="M 625 438 V 392" id="tfg-foresight-in"></path><path class="tf-link tf-link--rose tf-link--dashed tf-training-only" d="M 640 170 V 84" id="tfg-foresight-up"></path><path class="tf-link tf-link--teal" d="M 800 438 V 392" id="tfg-action-in"></path><path class="tf-link tf-link--teal" d="M 800 170 V 130" id="tfg-action-out"></path><g class="tf-dim"><rect class="tf-repeat-frame" height="134" rx="12" width="350" x="70" y="238"></rect><text class="tf-text--label tf-text--repeat" text-anchor="end" x="406" y="362">×6</text></g><g class="tf-dim tf-dim--expert"><rect class="tf-repeat-frame" height="134" rx="12" width="320" x="565" y="238"></rect><text class="tf-text--label tf-text--repeat" text-anchor="end" x="871" y="362">×6</text></g><g class="tf-dim"><rect class="tf-layer" height="42" rx="8" width="278" x="106" y="258"></rect><text class="tf-text" text-anchor="middle" x="245" y="285">1 × Full Attention</text></g><g class="tf-dim"><rect class="tf-layer" height="42" rx="8" width="278" x="106" y="316"></rect><text class="tf-text" text-anchor="middle" x="245" y="343">3 × Gated DeltaNet</text></g><g class="tf-dim tf-dim--expert"><rect class="tf-layer" height="42" rx="8" width="242" x="604" y="258"></rect><text class="tf-text" text-anchor="middle" x="725" y="285">1 × Full Attention</text></g><g class="tf-dim tf-dim--expert"><rect class="tf-layer" height="42" rx="8" width="242" x="604" y="316"></rect><text class="tf-text" text-anchor="middle" x="725" y="343">3 × Gated DeltaNet</text></g><text class="tf-text--label tf-text--mot" text-anchor="middle" x="498" y="269">MoT</text><text class="tf-text--label tf-training-only" text-anchor="middle" x="640" y="152">Foresight readout</text><g class="tf-flow-path"><path class="tf-flow-rail" d="M 890 113 H 942 Q 952 113 952 123 V 445 Q 952 455 942 455 H 890" id="tfg-flow-rail"></path></g><g class="tf-dim tf-training-only"><rect class="tf-box--wan" height="60" rx="10" width="220" x="545" y="24"></rect><text class="tf-text tf-text--strong" text-anchor="middle" x="655" y="49">Video Model</text><text class="tf-text--sub" text-anchor="middle" x="655" y="69">WAN2.2-5B · frozen</text></g><g class="tf-dim"><rect class="tf-pill" height="34" rx="17" width="140" x="90" y="96"></rect><text class="tf-text" text-anchor="middle" x="160" y="118">QA / SubTask</text></g><g class="tf-dim tf-training-only"><rect class="tf-pill" height="34" rx="17" width="170" x="310" y="96"></rect><text class="tf-text" text-anchor="middle" x="395" y="118">FAST Action Tokens</text></g><g class="tf-dim"><rect class="tf-pill" height="34" rx="17" width="180" x="710" y="96"></rect><text class="tf-text" text-anchor="middle" x="800" y="118">Action Chunk</text></g><g class="tf-dim is-active"><rect class="tf-pill" height="34" rx="17" width="88" x="48" y="438"></rect><text class="tf-text" text-anchor="middle" x="92" y="460">Images</text></g><g class="tf-dim is-active"><rect class="tf-pill" height="34" rx="17" width="108" x="146" y="438"></rect><text class="tf-text" text-anchor="middle" x="200" y="460">Instruction</text></g><g class="tf-dim is-active"><rect class="tf-pill" height="34" rx="17" width="72" x="264" y="438"></rect><text class="tf-text" text-anchor="middle" x="300" y="460">Mode</text></g><g class="tf-dim is-active"><rect class="tf-pill" height="34" rx="17" width="72" x="346" y="438"></rect><text class="tf-text" text-anchor="middle" x="382" y="460">State</text></g><g class="tf-dim"><rect class="tf-pill tf-pill--dashed" height="34" rx="17" width="150" x="550" y="438"></rect><text class="tf-text" text-anchor="middle" x="625" y="460">Foresight Tokens</text></g><g class="tf-dim"><rect class="tf-pill tf-pill--dashed" height="34" rx="17" width="180" x="710" y="438"></rect><text class="tf-text" text-anchor="middle" x="800" y="460">Noisy Action Queries</text></g><g class="tf-particles"><circle class="tf-token tf-token--vision" data-path-id="tfg-in-images" r="5" opacity="0" cx="0" cy="0"></circle><circle class="tf-token tf-token--language" data-path-id="tfg-in-instruction" r="5" opacity="0" cx="0" cy="0"></circle><circle class="tf-token tf-token--state" data-path-id="tfg-in-mode" r="5" opacity="0" cx="0" cy="0"></circle><circle class="tf-token tf-token--state" data-path-id="tfg-in-state" r="5" opacity="0" cx="0" cy="0"></circle></g></svg></div><div class="tf-narration"><div class="tf-chips"><button class="tf-chip is-active" type="button"><span>1</span>Tokenize</button><button class="tf-chip" type="button"><span>2</span>Model Forward</button><button class="tf-chip" type="button"><span>3</span>VLM Outputs</button><button class="tf-chip" type="button"><span>4</span>Foresight</button><button class="tf-chip" type="button"><span>5</span>Flow Matching</button></div><p aria-live="polite" class="tf-blurb"><strong>Tokenize the scene<!-- -->.</strong> <!-- -->Multi-view images, the language instruction, a control-mode token, and the 256-bin discretized robot state are embedded into one token sequence for the VLM.</p></div></div></figure></div></section><section class="border-t border-border px-5 py-18 sm:py-24 bg-muted/35" id="data"><div class="mx-auto max-w-[1120px]"><div class="mb-10 max-w-4xl"><p class="section-eyebrow mb-3">Data</p><h2 class="text-balance text-3xl font-semibold sm:text-5xl">1.2M robot episodes, 861M frames, and 3M multimodal samples</h2><p class="mt-5 text-lg leading-8 text-muted-foreground">A robot manipulation stream supplies action and future-frame supervision, while a multimodal stream preserves semantic and spatial-grounding ability.</p></div><figure class="overflow-hidden rounded-lg border border-border bg-card p-4 shadow-sm sm:p-6"><img alt="Training data overview for InternVLA-A1.5." loading="lazy" width="3375" height="1595" decoding="async" data-nimg="1" class="h-auto w-full rounded-md bg-white" style="color:transparent" src="/internvla-a15.github.io/reference-assets/paper/data-recipe.png"/><figcaption class="px-2 pb-1 pt-3 text-sm leading-6 text-muted-foreground">Training data overview: six robot manipulation sources plus an InternVLA-M1 multimodal corpus for VQA and robotics-oriented grounding.</figcaption></figure><div class="mt-8 grid gap-6"><figure class="overflow-hidden rounded-lg border border-border bg-card p-4 shadow-sm sm:p-6"><div class="results-table-shell"><table class="results-table"><thead><tr><th>Source</th><th>Type</th><th>Episodes</th><th>Frames</th><th>Weight</th></tr></thead><tbody><tr><td>InternData-A1</td><td>Sim.</td><td>587,946</td><td>395.9M</td><td>0.20</td></tr><tr><td>AgiBotWorld</td><td>Real</td><td>112,988</td><td>206.3M</td><td>0.25</td></tr><tr><td>UMI</td><td>Real</td><td>377,018</td><td>201.3M</td><td>0.10</td></tr><tr><td>DROID</td><td>Real</td><td>95,658</td><td>27.6M</td><td>0.15</td></tr><tr><td>Galaxea</td><td>Real</td><td>19,085</td><td>25.0M</td><td>0.20</td></tr><tr><td>RoboMind 1.0</td><td>Real</td><td>8,638</td><td>5.4M</td><td>0.10</td></tr></tbody></table></div><figcaption class="px-2 pb-1 pt-3 text-sm leading-6 text-muted-foreground">Robot manipulation data mixture.</figcaption></figure><figure class="overflow-hidden rounded-lg border border-border bg-card p-4 shadow-sm sm:p-6"><div class="results-table-shell"><table class="results-table"><thead><tr><th>Source</th><th>Samples</th><th>Role</th></tr></thead><tbody><tr><td>General QA</td><td>637K</td><td>Captioning, VQA, OCR, and knowledge grounding</td></tr><tr><td>Box QA</td><td>879K</td><td>Referring detection</td></tr><tr><td>Point QA</td><td>832K</td><td>Free-space and object-point localization</td></tr><tr><td>Trajectory QA</td><td>684K</td><td>End-effector waypoint prediction</td></tr></tbody></table></div><figcaption class="px-2 pb-1 pt-3 text-sm leading-6 text-muted-foreground">Multimodal co-training data from InternVLA-M1.</figcaption></figure></div></div></section><section class="border-t border-border px-5 py-18 sm:py-24" id="benchmarks"><div class="mx-auto max-w-[1120px]"><div class="mb-10 max-w-4xl"><p class="section-eyebrow mb-3">Benchmarks</p><h2 class="text-balance text-3xl font-semibold sm:text-5xl">Broad simulation gains across static, robust, dynamic, and long-horizon benchmarks</h2><p class="mt-5 text-lg leading-8 text-muted-foreground">The current paper reports simulation benchmark results across LIBERO, LIBERO-Plus, RoboTwin 2.0, SimplerEnv, DOMINO, and EBench.</p></div><figure class="overflow-hidden rounded-lg border border-border bg-card p-4 shadow-sm sm:p-6"><div class="results-table-shell"><table class="results-table results-summary-table"><thead><tr><th>Benchmark</th><th>Metric</th><th>InternVLA-A1.5</th></tr></thead><tbody><tr class="is-ours"><td><button aria-controls="results-detail-simplerenv" aria-expanded="false" class="results-expand-button" type="button"><svg xmlns="http://www.w3.org/2000/svg" width="24" height="24" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" class="lucide lucide-chevron-right results-expand-icon" aria-hidden="true"><path d="m9 18 6-6-6-6"></path></svg><span>SimplerEnv (WidowX)</span></button></td><td>Average SR</td><td>80.8</td></tr><tr class="is-ours"><td><button aria-controls="results-detail-robotwin" aria-expanded="false" class="results-expand-button" type="button"><svg xmlns="http://www.w3.org/2000/svg" width="24" height="24" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" class="lucide lucide-chevron-right results-expand-icon" aria-hidden="true"><path d="m9 18 6-6-6-6"></path></svg><span>RoboTwin 2.0</span></button></td><td>Average SR</td><td>93.2</td></tr><tr class="is-ours"><td><button aria-controls="results-detail-domino" aria-expanded="false" class="results-expand-button" type="button"><svg xmlns="http://www.w3.org/2000/svg" width="24" height="24" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" class="lucide lucide-chevron-right results-expand-icon" aria-hidden="true"><path d="m9 18 6-6-6-6"></path></svg><span>DOMINO</span></button></td><td>SR / MS</td><td>27.7 / 39.8</td></tr><tr class="is-ours"><td><button aria-controls="results-detail-libero" aria-expanded="false" class="results-expand-button" type="button"><svg xmlns="http://www.w3.org/2000/svg" width="24" height="24" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" class="lucide lucide-chevron-right results-expand-icon" aria-hidden="true"><path d="m9 18 6-6-6-6"></path></svg><span>LIBERO</span></button></td><td>Average SR</td><td>98.9</td></tr><tr class="is-ours"><td><button aria-controls="results-detail-libero-plus" aria-expanded="false" class="results-expand-button" type="button"><svg xmlns="http://www.w3.org/2000/svg" width="24" height="24" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" class="lucide lucide-chevron-right results-expand-icon" aria-hidden="true"><path d="m9 18 6-6-6-6"></path></svg><span>LIBERO-Plus</span></button></td><td>Average SR</td><td>84.8</td></tr><tr class="is-ours"><td><button aria-controls="results-detail-ebench" aria-expanded="false" class="results-expand-button" type="button"><svg xmlns="http://www.w3.org/2000/svg" width="24" height="24" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" class="lucide lucide-chevron-right results-expand-icon" aria-hidden="true"><path d="m9 18 6-6-6-6"></path></svg><span>EBench</span></button></td><td>SR / Score</td><td>35.2 / 49.5</td></tr></tbody></table></div><figcaption class="px-2 pb-1 pt-3 text-sm leading-6 text-muted-foreground">Click a benchmark row to view the full table.</figcaption></figure></div></section><section class="border-t border-border px-5 py-18 sm:py-24 bg-muted/35" id="experiments"><div class="mx-auto max-w-[1120px]"><div class="mb-10 max-w-4xl"><p class="section-eyebrow mb-3">Experiments</p><h2 class="text-balance text-3xl font-semibold sm:text-5xl">Instruction following and long-horizon chemistry tasks</h2><p class="mt-5 text-lg leading-8 text-muted-foreground">The draft defines a four-task real-world suite but does not yet contain finalized success-rate numbers, so this page only describes the evaluated task setup.</p></div><div class="grid gap-8"><figure class="overflow-hidden rounded-lg border border-border bg-card p-4 shadow-sm sm:p-6"><img alt="Four real-world tasks for InternVLA-A1.5 evaluation." loading="lazy" width="1722" height="366" decoding="async" data-nimg="1" class="h-auto w-full rounded-md bg-white" style="color:transparent" src="/internvla-a15.github.io/reference-assets/paper/realworld-tasks.png"/><figcaption class="px-2 pb-1 pt-3 text-sm leading-6 text-muted-foreground"><div class="real-world-caption"><p>The real-world suite includes Sort Tubes, Insert Tubes, Move Tubes, and MOF. Colored outlines mark task-relevant tubes and targets.</p><dl class="real-world-caption-grid"><div><dt>Sort Tubes</dt><dd>Instruction following with a specified tube and box target, including held-out arm-color bindings for compositional grounding.</dd></div><div><dt>Insert Tubes</dt><dd>The robot must bring a specified tube to a specified hole of a tube rack.</dd></div><div><dt>Move Tubes</dt><dd>The robot transports a specified tube to a specified position on the opposite side of the workspace.</dd></div><div><dt>MOF</dt><dd>A long-horizon chemistry procedure used to probe coherent multi-step execution.</dd></div></dl></div></figcaption></figure><figure class="overflow-hidden rounded-lg border border-border bg-card p-4 shadow-sm sm:p-6"><div class="realworld-video-showcase"><div class="realworld-video-header"><h3>Sort Tubes</h3><p class="realworld-video-description">Compositional Generalization: seen rollouts cover color and target factors; OOD rollouts swap them into held-out blue-right and orange-left instructions.</p></div><div aria-label="Seen blue plus left and seen orange plus right recombine into OOD blue plus right and OOD orange plus left." class="realworld-video-formula"><span>Seen: blue + left</span><span>Seen: orange + right</span><svg xmlns="http://www.w3.org/2000/svg" width="24" height="24" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" class="lucide lucide-arrow-right realworld-video-formula-icon" aria-hidden="true"><path d="M5 12h14"></path><path d="m12 5 7 7-7 7"></path></svg><span>OOD: blue + right</span><span>OOD: orange + left</span></div><div class="realworld-video-scroll"><div aria-label="Sort Tubes seen and OOD rollout matrix" class="realworld-video-matrix realworld-video-matrix--2"><div class="realworld-video-tile" data-split="Seen"><p class="realworld-video-tile-label"><strong>Seen<!-- -->:</strong><span>Blue tube</span><svg xmlns="http://www.w3.org/2000/svg" width="24" height="24" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" class="lucide lucide-arrow-right realworld-video-label-icon" aria-hidden="true"><path d="M5 12h14"></path><path d="m12 5 7 7-7 7"></path></svg><span>left box</span></p><video aria-label="Seen rollout: Blue tube to left box" autoPlay="" class="realworld-video" height="720" loop="" muted="" playsInline="" poster="/internvla-a15.github.io/videos/sorttube/seen-blue-left.jpg" preload="metadata" width="1280"><source src="/internvla-a15.github.io/videos/sorttube/seen-blue-left.mp4" type="video/mp4"/>Your browser does not support this video.<!-- --> <a href="/internvla-a15.github.io/videos/sorttube/seen-blue-left.mp4">Download this <!-- -->Sort Tubes<!-- --> rollout.</a></video></div><div class="realworld-video-tile" data-split="OOD"><p class="realworld-video-tile-label"><strong>OOD<!-- -->:</strong><span>Blue tube</span><svg xmlns="http://www.w3.org/2000/svg" width="24" height="24" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" class="lucide lucide-arrow-right realworld-video-label-icon" aria-hidden="true"><path d="M5 12h14"></path><path d="m12 5 7 7-7 7"></path></svg><span>right box</span></p><video aria-label="OOD rollout: Blue tube to right box" autoPlay="" class="realworld-video" height="720" loop="" muted="" playsInline="" poster="/internvla-a15.github.io/videos/sorttube/ood-blue-right.jpg" preload="metadata" width="1280"><source src="/internvla-a15.github.io/videos/sorttube/ood-blue-right.mp4" type="video/mp4"/>Your browser does not support this video.<!-- --> <a href="/internvla-a15.github.io/videos/sorttube/ood-blue-right.mp4">Download this <!-- -->Sort Tubes<!-- --> rollout.</a></video></div><div class="realworld-video-tile" data-split="OOD"><p class="realworld-video-tile-label"><strong>OOD<!-- -->:</strong><span>Orange tube</span><svg xmlns="http://www.w3.org/2000/svg" width="24" height="24" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" class="lucide lucide-arrow-right realworld-video-label-icon" aria-hidden="true"><path d="M5 12h14"></path><path d="m12 5 7 7-7 7"></path></svg><span>left box</span></p><video aria-label="OOD rollout: Orange tube to left box" autoPlay="" class="realworld-video" height="720" loop="" muted="" playsInline="" poster="/internvla-a15.github.io/videos/sorttube/ood-orange-left.jpg" preload="metadata" width="1280"><source src="/internvla-a15.github.io/videos/sorttube/ood-orange-left.mp4" type="video/mp4"/>Your browser does not support this video.<!-- --> <a href="/internvla-a15.github.io/videos/sorttube/ood-orange-left.mp4">Download this <!-- -->Sort Tubes<!-- --> rollout.</a></video></div><div class="realworld-video-tile" data-split="Seen"><p class="realworld-video-tile-label"><strong>Seen<!-- -->:</strong><span>Orange tube</span><svg xmlns="http://www.w3.org/2000/svg" width="24" height="24" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" class="lucide lucide-arrow-right realworld-video-label-icon" aria-hidden="true"><path d="M5 12h14"></path><path d="m12 5 7 7-7 7"></path></svg><span>right box</span></p><video aria-label="Seen rollout: Orange tube to right box" autoPlay="" class="realworld-video" height="720" loop="" muted="" playsInline="" poster="/internvla-a15.github.io/videos/sorttube/seen-orange-right.jpg" preload="metadata" width="1280"><source src="/internvla-a15.github.io/videos/sorttube/seen-orange-right.mp4" type="video/mp4"/>Your browser does not support this video.<!-- --> <a href="/internvla-a15.github.io/videos/sorttube/seen-orange-right.mp4">Download this <!-- -->Sort Tubes<!-- --> rollout.</a></video></div></div></div></div></figure><figure class="overflow-hidden rounded-lg border border-border bg-card p-4 shadow-sm sm:p-6"><div class="realworld-video-showcase"><div class="realworld-video-header"><h3>Insert Tubes</h3><p class="realworld-video-description">Compositional Generalization: the policy binds tube color to hole index and transfers that binding to held-out color-hole pairings.</p></div><div class="realworld-video-scroll"><div aria-label="Insert Tubes row-major rollout matrix" class="realworld-video-matrix realworld-video-matrix--4"><div class="realworld-video-tile" data-split="Seen"><p class="realworld-video-tile-label"><strong>Seen<!-- -->:</strong><span>Blue</span><svg xmlns="http://www.w3.org/2000/svg" width="24" height="24" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" class="lucide lucide-arrow-right realworld-video-label-icon" aria-hidden="true"><path d="M5 12h14"></path><path d="m12 5 7 7-7 7"></path></svg><span>Hole 1</span></p><video aria-label="Seen rollout: Blue tube to Hole 1" autoPlay="" class="realworld-video" height="720" loop="" muted="" playsInline="" poster="/internvla-a15.github.io/videos/inserttube/seen-b1.jpg" preload="metadata" width="1280"><source src="/internvla-a15.github.io/videos/inserttube/seen-b1.mp4" type="video/mp4"/>Your browser does not support this video.<!-- --> <a href="/internvla-a15.github.io/videos/inserttube/seen-b1.mp4">Download this <!-- -->Insert Tubes<!-- --> rollout.</a></video></div><div class="realworld-video-tile" data-split="OOD"><p class="realworld-video-tile-label"><strong>OOD<!-- -->:</strong><span>Blue</span><svg xmlns="http://www.w3.org/2000/svg" width="24" height="24" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" class="lucide lucide-arrow-right realworld-video-label-icon" aria-hidden="true"><path d="M5 12h14"></path><path d="m12 5 7 7-7 7"></path></svg><span>Hole 2</span></p><video aria-label="OOD rollout: Blue tube to Hole 2" autoPlay="" class="realworld-video" height="720" loop="" muted="" playsInline="" poster="/internvla-a15.github.io/videos/inserttube/ood-b2.jpg" preload="metadata" width="1280"><source src="/internvla-a15.github.io/videos/inserttube/ood-b2.mp4" type="video/mp4"/>Your browser does not support this video.<!-- --> <a href="/internvla-a15.github.io/videos/inserttube/ood-b2.mp4">Download this <!-- -->Insert Tubes<!-- --> rollout.</a></video></div><div class="realworld-video-tile" data-split="Seen"><p class="realworld-video-tile-label"><strong>Seen<!-- -->:</strong><span>Blue</span><svg xmlns="http://www.w3.org/2000/svg" width="24" height="24" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" class="lucide lucide-arrow-right realworld-video-label-icon" aria-hidden="true"><path d="M5 12h14"></path><path d="m12 5 7 7-7 7"></path></svg><span>Hole 3</span></p><video aria-label="Seen rollout: Blue tube to Hole 3" autoPlay="" class="realworld-video" height="720" loop="" muted="" playsInline="" poster="/internvla-a15.github.io/videos/inserttube/seen-b3.jpg" preload="metadata" width="1280"><source src="/internvla-a15.github.io/videos/inserttube/seen-b3.mp4" type="video/mp4"/>Your browser does not support this video.<!-- --> <a href="/internvla-a15.github.io/videos/inserttube/seen-b3.mp4">Download this <!-- -->Insert Tubes<!-- --> rollout.</a></video></div><div class="realworld-video-tile" data-split="Seen"><p class="realworld-video-tile-label"><strong>Seen<!-- -->:</strong><span>Blue</span><svg xmlns="http://www.w3.org/2000/svg" width="24" height="24" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" class="lucide lucide-arrow-right realworld-video-label-icon" aria-hidden="true"><path d="M5 12h14"></path><path d="m12 5 7 7-7 7"></path></svg><span>Hole 4</span></p><video aria-label="Seen rollout: Blue tube to Hole 4" autoPlay="" class="realworld-video" height="720" loop="" muted="" playsInline="" poster="/internvla-a15.github.io/videos/inserttube/seen-b4.jpg" preload="metadata" width="1280"><source src="/internvla-a15.github.io/videos/inserttube/seen-b4.mp4" type="video/mp4"/>Your browser does not support this video.<!-- --> <a href="/internvla-a15.github.io/videos/inserttube/seen-b4.mp4">Download this <!-- -->Insert Tubes<!-- --> rollout.</a></video></div><div class="realworld-video-tile" data-split="Seen"><p class="realworld-video-tile-label"><strong>Seen<!-- -->:</strong><span>Orange</span><svg xmlns="http://www.w3.org/2000/svg" width="24" height="24" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" class="lucide lucide-arrow-right realworld-video-label-icon" aria-hidden="true"><path d="M5 12h14"></path><path d="m12 5 7 7-7 7"></path></svg><span>Hole 1</span></p><video aria-label="Seen rollout: Orange tube to Hole 1" autoPlay="" class="realworld-video" height="720" loop="" muted="" playsInline="" poster="/internvla-a15.github.io/videos/inserttube/seen-o1.jpg" preload="metadata" width="1280"><source src="/internvla-a15.github.io/videos/inserttube/seen-o1.mp4" type="video/mp4"/>Your browser does not support this video.<!-- --> <a href="/internvla-a15.github.io/videos/inserttube/seen-o1.mp4">Download this <!-- -->Insert Tubes<!-- --> rollout.</a></video></div><div class="realworld-video-tile" data-split="Seen"><p class="realworld-video-tile-label"><strong>Seen<!-- -->:</strong><span>Orange</span><svg xmlns="http://www.w3.org/2000/svg" width="24" height="24" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" class="lucide lucide-arrow-right realworld-video-label-icon" aria-hidden="true"><path d="M5 12h14"></path><path d="m12 5 7 7-7 7"></path></svg><span>Hole 2</span></p><video aria-label="Seen rollout: Orange tube to Hole 2" autoPlay="" class="realworld-video" height="720" loop="" muted="" playsInline="" poster="/internvla-a15.github.io/videos/inserttube/seen-o2.jpg" preload="metadata" width="1280"><source src="/internvla-a15.github.io/videos/inserttube/seen-o2.mp4" type="video/mp4"/>Your browser does not support this video.<!-- --> <a href="/internvla-a15.github.io/videos/inserttube/seen-o2.mp4">Download this <!-- -->Insert Tubes<!-- --> rollout.</a></video></div><div class="realworld-video-tile" data-split="Seen"><p class="realworld-video-tile-label"><strong>Seen<!-- -->:</strong><span>Orange</span><svg xmlns="http://www.w3.org/2000/svg" width="24" height="24" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" class="lucide lucide-arrow-right realworld-video-label-icon" aria-hidden="true"><path d="M5 12h14"></path><path d="m12 5 7 7-7 7"></path></svg><span>Hole 3</span></p><video aria-label="Seen rollout: Orange tube to Hole 3" autoPlay="" class="realworld-video" height="720" loop="" muted="" playsInline="" poster="/internvla-a15.github.io/videos/inserttube/seen-o3.jpg" preload="metadata" width="1280"><source src="/internvla-a15.github.io/videos/inserttube/seen-o3.mp4" type="video/mp4"/>Your browser does not support this video.<!-- --> <a href="/internvla-a15.github.io/videos/inserttube/seen-o3.mp4">Download this <!-- -->Insert Tubes<!-- --> rollout.</a></video></div><div class="realworld-video-tile" data-split="OOD"><p class="realworld-video-tile-label"><strong>OOD<!-- -->:</strong><span>Orange</span><svg xmlns="http://www.w3.org/2000/svg" width="24" height="24" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" class="lucide lucide-arrow-right realworld-video-label-icon" aria-hidden="true"><path d="M5 12h14"></path><path d="m12 5 7 7-7 7"></path></svg><span>Hole 4</span></p><video aria-label="OOD rollout: Orange tube to Hole 4" autoPlay="" class="realworld-video" height="720" loop="" muted="" playsInline="" poster="/internvla-a15.github.io/videos/inserttube/ood-o4.jpg" preload="metadata" width="1280"><source src="/internvla-a15.github.io/videos/inserttube/ood-o4.mp4" type="video/mp4"/>Your browser does not support this video.<!-- --> <a href="/internvla-a15.github.io/videos/inserttube/ood-o4.mp4">Download this <!-- -->Insert Tubes<!-- --> rollout.</a></video></div></div></div></div></figure><figure class="overflow-hidden rounded-lg border border-border bg-card p-4 shadow-sm sm:p-6"><div class="realworld-video-showcase"><div class="realworld-video-header"><h3>Move Tubes</h3><p class="realworld-video-description">Compositional Generalization: the model reuses color and target-hole factors across a different manipulation primitive and OOD pairings.</p></div><div class="realworld-video-scroll"><div aria-label="Move Tubes row-major rollout matrix" class="realworld-video-matrix realworld-video-matrix--4"><div class="realworld-video-tile" data-split="OOD"><p class="realworld-video-tile-label"><strong>OOD<!-- -->:</strong><span>Blue</span><svg xmlns="http://www.w3.org/2000/svg" width="24" height="24" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" class="lucide lucide-arrow-right realworld-video-label-icon" aria-hidden="true"><path d="M5 12h14"></path><path d="m12 5 7 7-7 7"></path></svg><span>Hole 1</span></p><video aria-label="OOD rollout: Blue tube to Hole 1" autoPlay="" class="realworld-video" height="720" loop="" muted="" playsInline="" poster="/internvla-a15.github.io/videos/movetube/ood-b1.jpg" preload="metadata" width="1280"><source src="/internvla-a15.github.io/videos/movetube/ood-b1.mp4" type="video/mp4"/>Your browser does not support this video.<!-- --> <a href="/internvla-a15.github.io/videos/movetube/ood-b1.mp4">Download this <!-- -->Move Tubes<!-- --> rollout.</a></video></div><div class="realworld-video-tile" data-split="Seen"><p class="realworld-video-tile-label"><strong>Seen<!-- -->:</strong><span>Blue</span><svg xmlns="http://www.w3.org/2000/svg" width="24" height="24" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" class="lucide lucide-arrow-right realworld-video-label-icon" aria-hidden="true"><path d="M5 12h14"></path><path d="m12 5 7 7-7 7"></path></svg><span>Hole 2</span></p><video aria-label="Seen rollout: Blue tube to Hole 2" autoPlay="" class="realworld-video" height="720" loop="" muted="" playsInline="" poster="/internvla-a15.github.io/videos/movetube/seen-b2.jpg" preload="metadata" width="1280"><source src="/internvla-a15.github.io/videos/movetube/seen-b2.mp4" type="video/mp4"/>Your browser does not support this video.<!-- --> <a href="/internvla-a15.github.io/videos/movetube/seen-b2.mp4">Download this <!-- -->Move Tubes<!-- --> rollout.</a></video></div><div class="realworld-video-tile" data-split="OOD"><p class="realworld-video-tile-label"><strong>OOD<!-- -->:</strong><span>Blue</span><svg xmlns="http://www.w3.org/2000/svg" width="24" height="24" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" class="lucide lucide-arrow-right realworld-video-label-icon" aria-hidden="true"><path d="M5 12h14"></path><path d="m12 5 7 7-7 7"></path></svg><span>Hole 3</span></p><video aria-label="OOD rollout: Blue tube to Hole 3" autoPlay="" class="realworld-video" height="720" loop="" muted="" playsInline="" poster="/internvla-a15.github.io/videos/movetube/ood-b3.jpg" preload="metadata" width="1280"><source src="/internvla-a15.github.io/videos/movetube/ood-b3.mp4" type="video/mp4"/>Your browser does not support this video.<!-- --> <a href="/internvla-a15.github.io/videos/movetube/ood-b3.mp4">Download this <!-- -->Move Tubes<!-- --> rollout.</a></video></div><div class="realworld-video-tile" data-split="Seen"><p class="realworld-video-tile-label"><strong>Seen<!-- -->:</strong><span>Blue</span><svg xmlns="http://www.w3.org/2000/svg" width="24" height="24" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" class="lucide lucide-arrow-right realworld-video-label-icon" aria-hidden="true"><path d="M5 12h14"></path><path d="m12 5 7 7-7 7"></path></svg><span>Hole 4</span></p><video aria-label="Seen rollout: Blue tube to Hole 4" autoPlay="" class="realworld-video" height="720" loop="" muted="" playsInline="" poster="/internvla-a15.github.io/videos/movetube/seen-b4.jpg" preload="metadata" width="1280"><source src="/internvla-a15.github.io/videos/movetube/seen-b4.mp4" type="video/mp4"/>Your browser does not support this video.<!-- --> <a href="/internvla-a15.github.io/videos/movetube/seen-b4.mp4">Download this <!-- -->Move Tubes<!-- --> rollout.</a></video></div><div class="realworld-video-tile" data-split="Seen"><p class="realworld-video-tile-label"><strong>Seen<!-- -->:</strong><span>Orange</span><svg xmlns="http://www.w3.org/2000/svg" width="24" height="24" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" class="lucide lucide-arrow-right realworld-video-label-icon" aria-hidden="true"><path d="M5 12h14"></path><path d="m12 5 7 7-7 7"></path></svg><span>Hole 1</span></p><video aria-label="Seen rollout: Orange tube to Hole 1" autoPlay="" class="realworld-video" height="720" loop="" muted="" playsInline="" poster="/internvla-a15.github.io/videos/movetube/seen-o1.jpg" preload="metadata" width="1280"><source src="/internvla-a15.github.io/videos/movetube/seen-o1.mp4" type="video/mp4"/>Your browser does not support this video.<!-- --> <a href="/internvla-a15.github.io/videos/movetube/seen-o1.mp4">Download this <!-- -->Move Tubes<!-- --> rollout.</a></video></div><div class="realworld-video-tile" data-split="OOD"><p class="realworld-video-tile-label"><strong>OOD<!-- -->:</strong><span>Orange</span><svg xmlns="http://www.w3.org/2000/svg" width="24" height="24" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" class="lucide lucide-arrow-right realworld-video-label-icon" aria-hidden="true"><path d="M5 12h14"></path><path d="m12 5 7 7-7 7"></path></svg><span>Hole 2</span></p><video aria-label="OOD rollout: Orange tube to Hole 2" autoPlay="" class="realworld-video" height="720" loop="" muted="" playsInline="" poster="/internvla-a15.github.io/videos/movetube/ood-o2.jpg" preload="metadata" width="1280"><source src="/internvla-a15.github.io/videos/movetube/ood-o2.mp4" type="video/mp4"/>Your browser does not support this video.<!-- --> <a href="/internvla-a15.github.io/videos/movetube/ood-o2.mp4">Download this <!-- -->Move Tubes<!-- --> rollout.</a></video></div><div class="realworld-video-tile" data-split="Seen"><p class="realworld-video-tile-label"><strong>Seen<!-- -->:</strong><span>Orange</span><svg xmlns="http://www.w3.org/2000/svg" width="24" height="24" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" class="lucide lucide-arrow-right realworld-video-label-icon" aria-hidden="true"><path d="M5 12h14"></path><path d="m12 5 7 7-7 7"></path></svg><span>Hole 3</span></p><video aria-label="Seen rollout: Orange tube to Hole 3" autoPlay="" class="realworld-video" height="720" loop="" muted="" playsInline="" poster="/internvla-a15.github.io/videos/movetube/seen-o3.jpg" preload="metadata" width="1280"><source src="/internvla-a15.github.io/videos/movetube/seen-o3.mp4" type="video/mp4"/>Your browser does not support this video.<!-- --> <a href="/internvla-a15.github.io/videos/movetube/seen-o3.mp4">Download this <!-- -->Move Tubes<!-- --> rollout.</a></video></div><div class="realworld-video-tile" data-split="OOD"><p class="realworld-video-tile-label"><strong>OOD<!-- -->:</strong><span>Orange</span><svg xmlns="http://www.w3.org/2000/svg" width="24" height="24" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" class="lucide lucide-arrow-right realworld-video-label-icon" aria-hidden="true"><path d="M5 12h14"></path><path d="m12 5 7 7-7 7"></path></svg><span>Hole 4</span></p><video aria-label="OOD rollout: Orange tube to Hole 4" autoPlay="" class="realworld-video" height="720" loop="" muted="" playsInline="" poster="/internvla-a15.github.io/videos/movetube/ood-o4.jpg" preload="metadata" width="1280"><source src="/internvla-a15.github.io/videos/movetube/ood-o4.mp4" type="video/mp4"/>Your browser does not support this video.<!-- --> <a href="/internvla-a15.github.io/videos/movetube/ood-o4.mp4">Download this <!-- -->Move Tubes<!-- --> rollout.</a></video></div></div></div></div></figure></div></div></section><section class="border-t border-border px-5 py-18 sm:py-24 bg-muted/35" id="bibtex"><div class="mx-auto max-w-[1120px]"><div class="mb-10 max-w-4xl"><p class="section-eyebrow mb-3">Citation</p><h2 class="text-balance text-3xl font-semibold sm:text-5xl">BibTeX</h2></div><pre class="overflow-x-auto rounded-lg border border-border bg-card p-5 text-sm leading-7 text-foreground"><code>@misc{internvla_a15,
title = {InternVLA-A1.5: Unifying Understanding, Latent Foresight, and Action for Compositional Generalization},
author = {InternVLA-A1.5 team},
year = {2026},
howpublished = {\url{https://github.com/InternRobotics/InternVLA-A1.5}}
}</code></pre></div></section></main><!--$--><!--/$--><script src="/internvla-a15.github.io/_next/static/chunks/0emxhq679m-p2.js" id="_R_" async=""></script><script>(self.__next_f=self.__next_f||[]).push([0])</script><script>self.__next_f.push([1,"1:\"$Sreact.fragment\"\n2:I[27423,[\"/internvla-a15.github.io/_next/static/chunks/072b-ukyyvq~h.js\",\"/internvla-a15.github.io/_next/static/chunks/05.ce42pp9k.4.js\"],\"ThemeProvider\"]\n3:I[12224,[\"/internvla-a15.github.io/_next/static/chunks/072b-ukyyvq~h.js\",\"/internvla-a15.github.io/_next/static/chunks/05.ce42pp9k.4.js\"],\"default\"]\n4:I[67689,[\"/internvla-a15.github.io/_next/static/chunks/072b-ukyyvq~h.js\",\"/internvla-a15.github.io/_next/static/chunks/05.ce42pp9k.4.js\"],\"default\"]\n5:I[94013,[\"/internvla-a15.github.io/_next/static/chunks/072b-ukyyvq~h.js\",\"/internvla-a15.github.io/_next/static/chunks/05.ce42pp9k.4.js\",\"/internvla-a15.github.io/_next/static/chunks/03xdlv9wrei1r.js\"],\"ThemeToggle\"]\n14:I[77663,[\"/internvla-a15.github.io/_next/static/chunks/072b-ukyyvq~h.js\",\"/internvla-a15.github.io/_next/static/chunks/05.ce42pp9k.4.js\"],\"default\",1]\n:HL[\"/internvla-a15.github.io/_next/static/chunks/025_jc0.9ca-..css\",\"style\"]\n"])</script><script>self.__next_f.push([1,"0:{\"P\":null,\"c\":[\"\",\"\"],\"q\":\"\",\"i\":false,\"f\":[[[\"\",{\"children\":[\"__PAGE__\",{}]},\"$undefined\",\"$undefined\",16],[[\"$\",\"$1\",\"c\",{\"children\":[[[\"$\",\"link\",\"0\",{\"rel\":\"stylesheet\",\"href\":\"/internvla-a15.github.io/_next/static/chunks/025_jc0.9ca-..css\",\"precedence\":\"next\",\"crossOrigin\":\"$undefined\",\"nonce\":\"$undefined\"}],[\"$\",\"script\",\"script-0\",{\"src\":\"/internvla-a15.github.io/_next/static/chunks/072b-ukyyvq~h.js\",\"async\":true,\"nonce\":\"$undefined\"}],[\"$\",\"script\",\"script-1\",{\"src\":\"/internvla-a15.github.io/_next/static/chunks/05.ce42pp9k.4.js\",\"async\":true,\"nonce\":\"$undefined\"}]],[\"$\",\"html\",null,{\"lang\":\"en\",\"suppressHydrationWarning\":true,\"children\":[\"$\",\"body\",null,{\"children\":[\"$\",\"$L2\",null,{\"children\":[\"$\",\"$L3\",null,{\"parallelRouterKey\":\"children\",\"error\":\"$undefined\",\"errorStyles\":\"$undefined\",\"errorScripts\":\"$undefined\",\"template\":[\"$\",\"$L4\",null,{}],\"templateStyles\":\"$undefined\",\"templateScripts\":\"$undefined\",\"notFound\":[[[\"$\",\"title\",null,{\"children\":\"404: This page could not be found.\"}],[\"$\",\"div\",null,{\"style\":{\"fontFamily\":\"system-ui,\\\"Segoe UI\\\",Roboto,Helvetica,Arial,sans-serif,\\\"Apple Color Emoji\\\",\\\"Segoe UI Emoji\\\"\",\"height\":\"100vh\",\"textAlign\":\"center\",\"display\":\"flex\",\"flexDirection\":\"column\",\"alignItems\":\"center\",\"justifyContent\":\"center\"},\"children\":[\"$\",\"div\",null,{\"children\":[[\"$\",\"style\",null,{\"dangerouslySetInnerHTML\":{\"__html\":\"body{color:#000;background:#fff;margin:0}.next-error-h1{border-right:1px solid rgba(0,0,0,.3)}@media (prefers-color-scheme:dark){body{color:#fff;background:#000}.next-error-h1{border-right:1px solid rgba(255,255,255,.3)}}\"}}],[\"$\",\"h1\",null,{\"className\":\"next-error-h1\",\"style\":{\"display\":\"inline-block\",\"margin\":\"0 20px 0 0\",\"padding\":\"0 23px 0 0\",\"fontSize\":24,\"fontWeight\":500,\"verticalAlign\":\"top\",\"lineHeight\":\"49px\"},\"children\":404}],[\"$\",\"div\",null,{\"style\":{\"display\":\"inline-block\"},\"children\":[\"$\",\"h2\",null,{\"style\":{\"fontSize\":14,\"fontWeight\":400,\"lineHeight\":\"49px\",\"margin\":0},\"children\":\"This page could not be found.\"}]}]]}]}]],[]],\"forbidden\":\"$undefined\",\"unauthorized\":\"$undefined\"}]}]}]}]]}],{\"children\":[[\"$\",\"$1\",\"c\",{\"children\":[[[\"$\",\"header\",null,{\"className\":\"sticky top-0 z-40 border-b border-border/70 bg-background/82 backdrop-blur-xl\",\"children\":[\"$\",\"div\",null,{\"className\":\"mx-auto grid h-18 max-w-[1120px] grid-cols-[1fr_auto_1fr] items-center px-5\",\"children\":[[\"$\",\"a\",null,{\"className\":\"flex items-center gap-3 justify-self-start font-semibold\",\"href\":\"#top\",\"children\":[[\"$\",\"span\",null,{\"aria-hidden\":\"true\",\"className\":\"size-9 rounded-xl bg-[linear-gradient(135deg,#005ea8,#0d74ce_52%,#2aa8f2)] shadow-sm\"}],[\"$\",\"span\",null,{\"className\":\"text-lg\",\"children\":\"InternVLA-A1.5\"}]]}],[\"$\",\"nav\",null,{\"className\":\"hidden items-center gap-7 justify-self-center text-sm font-medium text-muted-foreground lg:flex\",\"children\":[[\"$\",\"a\",\"#introduction\",{\"className\":\"transition hover:text-foreground\",\"href\":\"#introduction\",\"children\":\"Introduction\"}],[\"$\",\"a\",\"#model\",{\"className\":\"transition hover:text-foreground\",\"href\":\"#model\",\"children\":\"Model\"}],[\"$\",\"a\",\"#data\",{\"className\":\"transition hover:text-foreground\",\"href\":\"#data\",\"children\":\"Data\"}],[\"$\",\"a\",\"#benchmarks\",{\"className\":\"transition hover:text-foreground\",\"href\":\"#benchmarks\",\"children\":\"Benchmarks\"}],[\"$\",\"a\",\"#experiments\",{\"className\":\"transition hover:text-foreground\",\"href\":\"#experiments\",\"children\":\"Experiments\"}],[\"$\",\"a\",\"#bibtex\",{\"className\":\"transition hover:text-foreground\",\"href\":\"#bibtex\",\"children\":\"BibTeX\"}]]}],[\"$\",\"div\",null,{\"className\":\"justify-self-end\",\"children\":[\"$\",\"$L5\",null,{}]}]]}]}],[\"$\",\"main\",null,{\"children\":[[\"$\",\"section\",null,{\"className\":\"hero-section px-5 pb-12 pt-14 sm:pt-18\",\"id\":\"top\",\"children\":[[\"$\",\"div\",null,{\"aria-hidden\":\"true\",\"className\":\"hero-background\"}],[\"$\",\"div\",null,{\"className\":\"mx-auto max-w-[1120px] text-center\",\"children\":[[\"$\",\"h1\",null,{\"className\":\"hero-title text-balance\",\"children\":[\"$\",\"span\",null,{\"className\":\"title-gradient\",\"children\":\"InternVLA-A1.5\"}]}],[\"$\",\"p\",null,{\"className\":\"hero-subtitle\",\"children\":[\"Unifying Understanding, Latent Foresight and\",[\"$\",\"br\",null,{}],\"Action for Compositional Generalization\"]}],[\"$\",\"div\",null,{\"className\":\"authors-line\",\"children\":[[\"$\",\"span\",\"Physical Intelligence Team, Shanghai AI Laboratory\",{\"className\":\"author-entry\",\"children\":[\"$\",\"strong\",null,{\"className\":\"author-name block\",\"children\":\"Physical Intelligence Team, Shanghai AI Laboratory\"}]}]]}],[\"$\",\"div\",null,{\"aria-label\":\"Affiliations\",\"className\":\"team-logo-row\",\"children\":[\"$L6\",\"$L7\"]}],\"$L8\",\"$L9\",\"$La\"]}]]}],\"$Lb\",\"$Lc\",\"$Ld\",\"$Le\",\"$Lf\",\"$L10\"]}]],[\"$L11\"],\"$L12\"]}],{},null,false,null]},null,false,null],\"$L13\",false]],\"m\":\"$undefined\",\"G\":[\"$14\",[\"$L15\"]],\"S\":true,\"h\":null,\"s\":\"$undefined\",\"l\":\"$undefined\",\"p\":\"$undefined\",\"d\":\"$undefined\",\"b\":\"1y91I-0nz9VO499cG07fc\"}\n"])</script><script>self.__next_f.push([1,"16:I[83540,[\"/internvla-a15.github.io/_next/static/chunks/072b-ukyyvq~h.js\",\"/internvla-a15.github.io/_next/static/chunks/05.ce42pp9k.4.js\",\"/internvla-a15.github.io/_next/static/chunks/03xdlv9wrei1r.js\"],\"Image\"]\n17:I[66990,[\"/internvla-a15.github.io/_next/static/chunks/072b-ukyyvq~h.js\",\"/internvla-a15.github.io/_next/static/chunks/05.ce42pp9k.4.js\",\"/internvla-a15.github.io/_next/static/chunks/03xdlv9wrei1r.js\"],\"default\"]\n19:I[90793,[\"/internvla-a15.github.io/_next/static/chunks/072b-ukyyvq~h.js\",\"/internvla-a15.github.io/_next/static/chunks/05.ce42pp9k.4.js\",\"/internvla-a15.github.io/_next/static/chunks/03xdlv9wrei1r.js\"],\"TransformerFlow\"]\n20:I[88565,[\"/internvla-a15.github.io/_next/static/chunks/072b-ukyyvq~h.js\",\"/internvla-a15.github.io/_next/static/chunks/05.ce42pp9k.4.js\",\"/internvla-a15.github.io/_next/static/chunks/03xdlv9wrei1r.js\"],\"ExpandableResultsTable\"]\n29:I[76779,[\"/internvla-a15.github.io/_next/static/chunks/072b-ukyyvq~h.js\",\"/internvla-a15.github.io/_next/static/chunks/05.ce42pp9k.4.js\"],\"OutletBoundary\"]\n2a:\"$Sreact.suspense\"\n2c:I[76779,[\"/internvla-a15.github.io/_next/static/chunks/072b-ukyyvq~h.js\",\"/internvla-a15.github.io/_next/static/chunks/05.ce42pp9k.4.js\"],\"ViewportBoundary\"]\n2e:I[76779,[\"/internvla-a15.github.io/_next/static/chunks/072b-ukyyvq~h.js\",\"/internvla-a15.github.io/_next/static/chunks/05.ce42pp9k.4.js\"],\"MetadataBoundary\"]\n6:[\"$\",\"$L16\",\"Shanghai Artificial Intelligence Laboratory\",{\"alt\":\"Shanghai Artificial Intelligence Laboratory\",\"className\":\"team-logo-image\",\"height\":390,\"src\":\"/internvla-a15.github.io/reference-assets/images/ailab-logo.png\",\"width\":672}]\n7:[\"$\",\"$L16\",\"InternRobotics\",{\"alt\":\"InternRobotics\",\"className\":\"team-logo-image\",\"height\":110,\"src\":\"/internvla-a15.github.io/reference-assets/images/InternRobotics-logo.png\",\"width\":739}]\n"])</script><script>self.__next_f.push([1,"8:[\"$\",\"div\",null,{\"className\":\"mx-auto mt-6 flex max-w-4xl flex-wrap justify-center gap-3\",\"children\":[[\"$\",\"a\",\"Paper\",{\"className\":\"inline-flex shrink-0 items-center justify-center gap-2 whitespace-nowrap rounded-full border text-sm font-medium transition-all outline-none focus-visible:ring-3 focus-visible:ring-ring/30 disabled:pointer-events-none disabled:opacity-50 [\u0026_svg]:pointer-events-none [\u0026_svg]:size-4 [\u0026_svg]:shrink-0 active:translate-y-px border-transparent bg-primary text-primary-foreground shadow-sm hover:bg-primary/90 h-10 px-5\",\"href\":\"/internvla-a15.github.io/reference-assets/paper/InternVLA_A1_5.pdf\",\"children\":[[\"$\",\"$L17\",null,{\"ref\":\"$undefined\",\"iconNode\":[[\"path\",{\"d\":\"M6 22a2 2 0 0 1-2-2V4a2 2 0 0 1 2-2h8a2.4 2.4 0 0 1 1.704.706l3.588 3.588A2.4 2.4 0 0 1 20 8v12a2 2 0 0 1-2 2z\",\"key\":\"1oefj6\"}],[\"path\",{\"d\":\"M14 2v5a1 1 0 0 0 1 1h5\",\"key\":\"wfsgrz\"}],[\"path\",{\"d\":\"M10 9H8\",\"key\":\"b1mrlr\"}],[\"path\",{\"d\":\"M16 13H8\",\"key\":\"t4e002\"}],[\"path\",{\"d\":\"M16 17H8\",\"key\":\"z1uh3a\"}]],\"className\":\"lucide-file-text\",\"aria-hidden\":\"true\"}],\"Paper\"]}],[\"$\",\"a\",\"Github\",{\"className\":\"inline-flex shrink-0 items-center justify-center gap-2 whitespace-nowrap rounded-full border text-sm font-medium transition-all outline-none focus-visible:ring-3 focus-visible:ring-ring/30 disabled:pointer-events-none disabled:opacity-50 [\u0026_svg]:pointer-events-none [\u0026_svg]:size-4 [\u0026_svg]:shrink-0 active:translate-y-px border-border/80 bg-card/70 text-foreground backdrop-blur-md hover:bg-muted h-10 px-5\",\"href\":\"https://github.com/InternRobotics/InternVLA-A-series\",\"children\":[[\"$\",\"$L16\",null,{\"alt\":\"\",\"aria-hidden\":\"true\",\"className\":\"size-4 shrink-0 object-contain invert dark:invert-0\",\"height\":16,\"src\":\"/internvla-a15.github.io/reference-assets/images/github-invertocat-white.svg\",\"width\":16}],\"Github\"]}],[\"$\",\"a\",\"Huggingface\",{\"className\":\"inline-flex shrink-0 items-center justify-center gap-2 whitespace-nowrap rounded-full border text-sm font-medium transition-all outline-none focus-visible:ring-3 focus-visible:ring-ring/30 disabled:pointer-events-none disabled:opacity-50 [\u0026_svg]:pointer-events-none [\u0026_svg]:size-4 [\u0026_svg]:shrink-0 active:translate-y-px border-border/80 bg-card/70 text-foreground backdrop-blur-md hover:bg-muted h-10 px-5\",\"href\":\"https://huggingface.co/collections/InternRobotics/internvla-a15\",\"children\":[[\"$\",\"$L16\",null,{\"alt\":\"\",\"aria-hidden\":\"true\",\"className\":\"size-4 shrink-0 object-contain\",\"height\":16,\"src\":\"/internvla-a15.github.io/reference-assets/images/huggingface-logo.svg\",\"width\":16}],\"Huggingface\"]}],[\"$\",\"a\",\"ModelScope\",{\"className\":\"inline-flex shrink-0 items-center justify-center gap-2 whitespace-nowrap rounded-full border text-sm font-medium transition-all outline-none focus-visible:ring-3 focus-visible:ring-ring/30 disabled:pointer-events-none disabled:opacity-50 [\u0026_svg]:pointer-events-none [\u0026_svg]:size-4 [\u0026_svg]:shrink-0 active:translate-y-px border-border/80 bg-card/70 text-foreground backdrop-blur-md hover:bg-muted h-10 px-5\",\"href\":\"https://www.modelscope.cn/collections/InternRobotics/InternVLA-A15\",\"children\":[[\"$\",\"$L16\",null,{\"alt\":\"\",\"aria-hidden\":\"true\",\"className\":\"size-4 shrink-0 object-contain\",\"height\":16,\"src\":\"/internvla-a15.github.io/reference-assets/images/modelscope-logo.svg\",\"width\":16}],\"ModelScope\"]}],[\"$\",\"a\",\"BibTeX\",{\"className\":\"inline-flex shrink-0 items-center justify-center gap-2 whitespace-nowrap rounded-full border text-sm font-medium transition-all outline-none focus-visible:ring-3 focus-visible:ring-ring/30 disabled:pointer-events-none disabled:opacity-50 [\u0026_svg]:pointer-events-none [\u0026_svg]:size-4 [\u0026_svg]:shrink-0 active:translate-y-px border-border/80 bg-card/70 text-foreground backdrop-blur-md hover:bg-muted h-10 px-5\",\"href\":\"#bibtex\",\"children\":[\"$L18\",\"BibTeX\"]}]]}]\n"])</script><script>self.__next_f.push([1,"9:[\"$\",\"figure\",null,{\"className\":\"hero-teaser\",\"children\":[[\"$\",\"$L16\",null,{\"alt\":\"Overview of InternVLA-A1.5, combining vision-language understanding, latent visual foresight, and action generation.\",\"className\":\"h-auto w-full rounded-md bg-white\",\"height\":1493,\"priority\":true,\"src\":\"/internvla-a15.github.io/reference-assets/paper/teaser.png\",\"width\":2320}],[\"$\",\"figcaption\",null,{\"children\":\"InternVLA-A1.5 uses learnable foresight tokens to query future-relevant latent information from shared multimodal context, while action queries decode continuous control through a lightweight expert.\"}]]}]\na:[\"$\",\"div\",null,{\"className\":\"stat-band\",\"children\":[[\"$\",\"div\",\"LIBERO average\",{\"className\":\"stat-card\",\"children\":[[\"$\",\"p\",null,{\"children\":\"98.9\"}],[\"$\",\"h2\",null,{\"children\":\"LIBERO average\"}],[\"$\",\"span\",null,{\"children\":\"Best average success rate on the four LIBERO suites.\"}]]}],[\"$\",\"div\",\"LIBERO-Plus total\",{\"className\":\"stat-card\",\"children\":[[\"$\",\"p\",null,{\"children\":\"84.8\"}],[\"$\",\"h2\",null,{\"children\":\"LIBERO-Plus total\"}],[\"$\",\"span\",null,{\"children\":\"Best total robustness under visual and layout perturbations.\"}]]}],[\"$\",\"div\",\"RoboTwin average\",{\"className\":\"stat-card\",\"children\":[[\"$\",\"p\",null,{\"children\":\"93.2\"}],[\"$\",\"h2\",null,{\"children\":\"RoboTwin average\"}],[\"$\",\"span\",null,{\"children\":\"Strong bimanual manipulation performance across clean and randomized splits.\"}]]}],[\"$\",\"div\",\"Training frames\",{\"className\":\"stat-card\",\"children\":[[\"$\",\"p\",null,{\"children\":\"861M\"}],[\"$\",\"h2\",null,{\"children\":\"Training frames\"}],[\"$\",\"span\",null,{\"children\":\"Robot manipulation corpus spanning 1.2M episodes.\"}]]}]]}]\nb:[\"$\",\"section\",null,{\"className\":\"border-t border-border px-5 py-18 sm:py-24\",\"id\":\"introduction\",\"children\":[\"$\",\"div\",null,{\"className\":\"mx-auto max-w-[1120px]\",\"children\":[[\"$\",\"div\",null,{\"className\":\"mb-10 max-w-4xl\",\"children\":[[\"$\",\"p\",null,{\"className\":\"section-eyebrow mb-3\",\"children\":\"Introduction\"}],[\"$\",\"h2\",null,{\"className\":\"text-balance text-3xl font-semibold sm:text-5xl\",\"children\":\"A Quick Overview of InternVLA-A1.5\"}],[\"$\",\"p\",null,{\"className\":\"mt-5 text-lg leading-8 text-muted-foreground\",\"children\":\"Start with the project video before the architecture walkthrough.\"}]]}],[\"$\",\"figure\",null,{\"className\":\"overflow-hidden rounded-lg border border-border bg-card p-4 shadow-sm sm:p-6\",\"children\":[[\"$\",\"video\",null,{\"className\":\"intro-video\",\"controls\":true,\"height\":1080,\"playsInline\":true,\"poster\":\"/internvla-a15.github.io/videos/intro-poster.jpg\",\"preload\":\"metadata\",\"width\":1920,\"children\":[[\"$\",\"source\",null,{\"src\":\"/internvla-a15.github.io/videos/intro.mp4\",\"type\":\"video/mp4\"}],\"Your browser does not support this video.\",\" \",[\"$\",\"a\",null,{\"href\":\"/internvla-a15.github.io/videos/intro.mp4\",\"children\":\"Download the introduction video.\"}]]}],null]}]]}]}]\nc:[\"$\",\"section\",null,{\"className\":\"border-t border-border px-5 py-18 sm:py-24 bg-muted/35\",\"id\":\"model\",\"children\":[\"$\",\"div\",null,{\"className\":\"mx-auto max-w-[1120px]\",\"children\":[[\"$\",\"div\",null,{\"className\":\"mb-10 max-w-4xl\",\"children\":[[\"$\",\"p\",null,{\"className\":\"section-eyebrow mb-3\",\"children\":\"Model\"}],[\"$\",\"h2\",null,{\"className\":\"text-balance text-3xl font-semibold sm:text-5xl\",\"children\":\"Mixture-of-Transformers with training-only latent foresight\"}],[\"$\",\"p\",null,{\"className\":\"mt-5 text-lg leading-8 text-muted-foreground\",\"children\":\"The model adopts a Mixture-of-Transformers design with a native VLM backbone and a smaller unified expert that receives multimodal context through shared full attention.\"}]]}],[\"$\",\"figure\",null,{\"className\":\"overflow-hidden rounded-lg border border-border bg-card p-4 shadow-sm sm:p-6\",\"children\":[[\"$\",\"$L19\",null,{}],null]}]]}]}]\n"])</script><script>self.__next_f.push([1,"d:[\"$\",\"section\",null,{\"className\":\"border-t border-border px-5 py-18 sm:py-24 bg-muted/35\",\"id\":\"data\",\"children\":[\"$\",\"div\",null,{\"className\":\"mx-auto max-w-[1120px]\",\"children\":[[\"$\",\"div\",null,{\"className\":\"mb-10 max-w-4xl\",\"children\":[[\"$\",\"p\",null,{\"className\":\"section-eyebrow mb-3\",\"children\":\"Data\"}],[\"$\",\"h2\",null,{\"className\":\"text-balance text-3xl font-semibold sm:text-5xl\",\"children\":\"1.2M robot episodes, 861M frames, and 3M multimodal samples\"}],[\"$\",\"p\",null,{\"className\":\"mt-5 text-lg leading-8 text-muted-foreground\",\"children\":\"A robot manipulation stream supplies action and future-frame supervision, while a multimodal stream preserves semantic and spatial-grounding ability.\"}]]}],[[\"$\",\"figure\",null,{\"className\":\"overflow-hidden rounded-lg border border-border bg-card p-4 shadow-sm sm:p-6\",\"children\":[[\"$\",\"$L16\",null,{\"alt\":\"Training data overview for InternVLA-A1.5.\",\"className\":\"h-auto w-full rounded-md bg-white\",\"height\":1595,\"priority\":\"$undefined\",\"src\":\"/internvla-a15.github.io/reference-assets/paper/data-recipe.png\",\"width\":3375}],[\"$\",\"figcaption\",null,{\"className\":\"px-2 pb-1 pt-3 text-sm leading-6 text-muted-foreground\",\"children\":\"Training data overview: six robot manipulation sources plus an InternVLA-M1 multimodal corpus for VQA and robotics-oriented grounding.\"}]]}],[\"$\",\"div\",null,{\"className\":\"mt-8 grid gap-6\",\"children\":[[\"$\",\"figure\",null,{\"className\":\"overflow-hidden rounded-lg border border-border bg-card p-4 shadow-sm sm:p-6\",\"children\":[[\"$\",\"div\",null,{\"className\":\"results-table-shell\",\"children\":[\"$\",\"table\",null,{\"className\":\"results-table\",\"children\":[[\"$\",\"thead\",null,{\"children\":[\"$\",\"tr\",null,{\"children\":[[\"$\",\"th\",\"Source\",{\"children\":\"Source\"}],[\"$\",\"th\",\"Type\",{\"children\":\"Type\"}],[\"$\",\"th\",\"Episodes\",{\"children\":\"Episodes\"}],[\"$\",\"th\",\"Frames\",{\"children\":\"Frames\"}],[\"$\",\"th\",\"Weight\",{\"children\":\"Weight\"}]]}]}],[\"$\",\"tbody\",null,{\"children\":[[\"$\",\"tr\",\"InternData-A1-0\",{\"className\":\"$undefined\",\"children\":[[\"$\",\"td\",\"InternData-A1-0\",{\"children\":\"InternData-A1\"}],[\"$\",\"td\",\"Sim.-1\",{\"children\":\"Sim.\"}],[\"$\",\"td\",\"587,946-2\",{\"children\":\"587,946\"}],[\"$\",\"td\",\"395.9M-3\",{\"children\":\"395.9M\"}],[\"$\",\"td\",\"0.20-4\",{\"children\":\"0.20\"}]]}],[\"$\",\"tr\",\"AgiBotWorld-1\",{\"className\":\"$undefined\",\"children\":[[\"$\",\"td\",\"AgiBotWorld-0\",{\"children\":\"AgiBotWorld\"}],[\"$\",\"td\",\"Real-1\",{\"children\":\"Real\"}],[\"$\",\"td\",\"112,988-2\",{\"children\":\"112,988\"}],[\"$\",\"td\",\"206.3M-3\",{\"children\":\"206.3M\"}],[\"$\",\"td\",\"0.25-4\",{\"children\":\"0.25\"}]]}],[\"$\",\"tr\",\"UMI-2\",{\"className\":\"$undefined\",\"children\":[[\"$\",\"td\",\"UMI-0\",{\"children\":\"UMI\"}],[\"$\",\"td\",\"Real-1\",{\"children\":\"Real\"}],[\"$\",\"td\",\"377,018-2\",{\"children\":\"377,018\"}],[\"$\",\"td\",\"201.3M-3\",{\"children\":\"201.3M\"}],[\"$\",\"td\",\"0.10-4\",{\"children\":\"0.10\"}]]}],[\"$\",\"tr\",\"DROID-3\",{\"className\":\"$undefined\",\"children\":[[\"$\",\"td\",\"DROID-0\",{\"children\":\"DROID\"}],[\"$\",\"td\",\"Real-1\",{\"children\":\"Real\"}],[\"$\",\"td\",\"95,658-2\",{\"children\":\"95,658\"}],[\"$\",\"td\",\"27.6M-3\",{\"children\":\"27.6M\"}],[\"$\",\"td\",\"0.15-4\",{\"children\":\"0.15\"}]]}],[\"$\",\"tr\",\"Galaxea-4\",{\"className\":\"$undefined\",\"children\":[[\"$\",\"td\",\"Galaxea-0\",{\"children\":\"Galaxea\"}],[\"$\",\"td\",\"Real-1\",{\"children\":\"Real\"}],[\"$\",\"td\",\"19,085-2\",{\"children\":\"19,085\"}],[\"$\",\"td\",\"25.0M-3\",{\"children\":\"25.0M\"}],[\"$\",\"td\",\"0.20-4\",{\"children\":\"0.20\"}]]}],[\"$\",\"tr\",\"RoboMind 1.0-5\",{\"className\":\"$undefined\",\"children\":[[\"$\",\"td\",\"RoboMind 1.0-0\",{\"children\":\"RoboMind 1.0\"}],[\"$\",\"td\",\"Real-1\",{\"children\":\"Real\"}],[\"$\",\"td\",\"8,638-2\",{\"children\":\"8,638\"}],[\"$\",\"td\",\"5.4M-3\",{\"children\":\"5.4M\"}],[\"$\",\"td\",\"0.10-4\",{\"children\":\"0.10\"}]]}]]}]]}]}],[\"$\",\"figcaption\",null,{\"className\":\"px-2 pb-1 pt-3 text-sm leading-6 text-muted-foreground\",\"children\":\"Robot manipulation data mixture.\"}]]}],[\"$\",\"figure\",null,{\"className\":\"overflow-hidden rounded-lg border border-border bg-card p-4 shadow-sm sm:p-6\",\"children\":[[\"$\",\"div\",null,{\"className\":\"results-table-shell\",\"children\":[\"$\",\"table\",null,{\"className\":\"results-table\",\"children\":[[\"$\",\"thead\",null,{\"children\":[\"$\",\"tr\",null,{\"children\":[[\"$\",\"th\",\"Source\",{\"children\":\"Source\"}],[\"$\",\"th\",\"Samples\",{\"children\":\"Samples\"}],[\"$\",\"th\",\"Role\",{\"children\":\"Role\"}]]}]}],[\"$\",\"tbody\",null,{\"children\":[[\"$\",\"tr\",\"General QA-0\",{\"className\":\"$undefined\",\"children\":[[\"$\",\"td\",\"General QA-0\",{\"children\":\"General QA\"}],[\"$\",\"td\",\"637K-1\",{\"children\":\"637K\"}],[\"$\",\"td\",\"Captioning, VQA, OCR, and knowledge grounding-2\",{\"children\":\"Captioning, VQA, OCR, and knowledge grounding\"}]]}],[\"$\",\"tr\",\"Box QA-1\",{\"className\":\"$undefined\",\"children\":[\"$L1a\",\"$L1b\",\"$L1c\"]}],\"$L1d\",\"$L1e\"]}]]}]}],\"$L1f\"]}]]}]]]}]}]\n"])</script><script>self.__next_f.push([1,"e:[\"$\",\"section\",null,{\"className\":\"border-t border-border px-5 py-18 sm:py-24\",\"id\":\"benchmarks\",\"children\":[\"$\",\"div\",null,{\"className\":\"mx-auto max-w-[1120px]\",\"children\":[[\"$\",\"div\",null,{\"className\":\"mb-10 max-w-4xl\",\"children\":[[\"$\",\"p\",null,{\"className\":\"section-eyebrow mb-3\",\"children\":\"Benchmarks\"}],[\"$\",\"h2\",null,{\"className\":\"text-balance text-3xl font-semibold sm:text-5xl\",\"children\":\"Broad simulation gains across static, robust, dynamic, and long-horizon benchmarks\"}],[\"$\",\"p\",null,{\"className\":\"mt-5 text-lg leading-8 text-muted-foreground\",\"children\":\"The current paper reports simulation benchmark results across LIBERO, LIBERO-Plus, RoboTwin 2.0, SimplerEnv, DOMINO, and EBench.\"}]]}],[\"$\",\"figure\",null,{\"className\":\"overflow-hidden rounded-lg border border-border bg-card p-4 shadow-sm sm:p-6\",\"children\":[[\"$\",\"$L20\",null,{\"results\":[{\"id\":\"simplerenv\",\"benchmark\":\"SimplerEnv (WidowX)\",\"metric\":\"Average SR\",\"value\":\"80.8\",\"detailCaption\":\"Visual-matching success rates on the WidowX Bridge setup.\",\"detailColumns\":[\"Method\",\"Spoon\",\"Carrot\",\"Stack block\",\"Eggplant\",\"Average\"],\"detailRows\":[{\"cells\":[\"pi-0\",\"29.1\",\"0.0\",\"16.6\",\"62.5\",\"27.1\"]},{\"cells\":[\"pi-0.5\",\"49.3\",\"64.7\",\"44.7\",\"69.7\",\"57.1\"]},{\"cells\":[\"GR00T-N1.5\",\"75.3\",\"54.3\",\"57.0\",\"61.3\",\"61.9\"]},{\"cells\":[\"InternVLA-M1\",\"87.5\",\"67.9\",\"31.3\",\"100.0\",\"71.7\"]},{\"cells\":[\"EO-1\",\"63.6\",\"54.5\",\"81.8\",\"90.9\",\"72.7\"]},{\"cells\":[\"Xiaomi-Robotics-0\",\"95.8\",\"62.5\",\"75.0\",\"83.3\",\"79.2\"]},{\"cells\":[\"InternVLA-A1.5 (Ours)\",\"92.4\",\"85.4\",\"69.5\",\"75.7\",\"80.8\"],\"highlight\":true}]},{\"id\":\"robotwin\",\"benchmark\":\"RoboTwin 2.0\",\"metric\":\"Average SR\",\"value\":\"93.2\",\"detailCaption\":\"Success rates under clean and randomized bimanual manipulation settings.\",\"detailColumns\":[\"Method\",\"Clean\",\"Rand.\",\"Average\"],\"detailRows\":[{\"cells\":[\"pi-0\",\"65.9\",\"58.4\",\"62.2\"]},{\"cells\":[\"pi-0.5\",\"82.7\",\"76.8\",\"79.8\"]},{\"cells\":[\"LingBot-VLA\",\"86.5\",\"85.3\",\"85.9\"]},{\"cells\":[\"StarVLA\",\"88.7\",\"87.8\",\"88.3\"]},{\"cells\":[\"InternVLA-A1\",\"89.4\",\"89.6\",\"89.5\"]},{\"cells\":[\"Motus\",\"88.7\",\"87.0\",\"87.8\"]},{\"cells\":[\"LingBot-VA\",\"92.9\",\"91.5\",\"92.2\"]},{\"cells\":[\"Fast-WAM\",\"91.9\",\"91.8\",\"91.8\"]},{\"cells\":[\"Being-H0.7\",\"90.2\",\"89.6\",\"89.9\"]},{\"cells\":[\"InternVLA-A1.5 (Ours)\",\"93.3\",\"93.0\",\"93.2\"],\"highlight\":true}]},{\"id\":\"domino\",\"benchmark\":\"DOMINO\",\"metric\":\"SR / MS\",\"value\":\"27.7 / 39.8\",\"detailCaption\":\"Zero-shot dynamic manipulation results; SR is the primary metric and MS is manipulation score.\",\"detailColumns\":[\"Method\",\"SR\",\"MS\"],\"detailRows\":[{\"cells\":[\"OpenVLA-OFT\",\"6.7\",\"20.0\"]},{\"cells\":[\"pi-0.5\",\"7.5\",\"20.4\"]},{\"cells\":[\"LingBot-VLA w/ depth\",\"11.8\",\"26.7\"]},{\"cells\":[\"LingBot-VA\",\"24.1\",\"36.1\"]},{\"cells\":[\"Qwen-VLA-Base\",\"21.1\",\"37.4\"]},{\"cells\":[\"Qwen-VLA-Instruct\",\"26.6\",\"39.5\"]},{\"cells\":[\"InternVLA-A1.5 (Zero-shot)\",\"27.7\",\"39.8\"],\"highlight\":true}]},{\"id\":\"libero\",\"benchmark\":\"LIBERO\",\"metric\":\"Average SR\",\"value\":\"98.9\",\"detailCaption\":\"Success rates on LIBERO Spatial, Object, Goal, and Long suites.\",\"detailColumns\":[\"Method\",\"Spatial\",\"Object\",\"Goal\",\"Long\",\"Average\"],\"detailRows\":[{\"cells\":[\"pi-0\",\"98.0\",\"96.8\",\"94.4\",\"88.4\",\"94.4\"]},{\"cells\":[\"pi-0.5\",\"98.8\",\"98.2\",\"98.0\",\"92.4\",\"96.9\"]},{\"cells\":[\"GR00T-N1.7\",\"97.7\",\"98.5\",\"97.5\",\"94.4\",\"97.0\"]},{\"cells\":[\"OpenVLA-OFT\",\"97.6\",\"98.4\",\"97.9\",\"94.5\",\"97.1\"]},{\"cells\":[\"InternVLA-M1\",\"98.0\",\"99.0\",\"93.8\",\"92.6\",\"95.9\"]},{\"cells\":[\"Xiaomi-Robotics-0\",\"98.8\",\"100.0\",\"98.8\",\"97.2\",\"98.7\"]},{\"cells\":[\"Motus\",\"96.8\",\"99.8\",\"96.6\",\"97.6\",\"97.7\"]},{\"cells\":[\"LingBot-VA\",\"98.5\",\"99.6\",\"97.2\",\"98.5\",\"98.5\"]},{\"cells\":[\"Fast-WAM\",\"98.2\",\"100.0\",\"97.0\",\"95.2\",\"97.6\"]},{\"cells\":[\"InternVLA-A1.5 (Ours)\",\"98.6\",\"99.8\",\"98.6\",\"98.4\",\"98.9\"],\"highlight\":true}]},{\"id\":\"libero-plus\",\"benchmark\":\"LIBERO-Plus\",\"metric\":\"Average SR\",\"value\":\"84.8\",\"detailCaption\":\"Zero-shot robustness under visual and layout perturbations.\",\"detailColumns\":[\"Method\",\"Camera\",\"Robot\",\"Language\",\"Light\",\"Background\",\"Noise\",\"Layout\",\"Average\"],\"detailRows\":[{\"cells\":[\"pi-0\",\"13.8\",\"6.0\",\"58.8\",\"85.0\",\"81.4\",\"79.0\",\"68.9\",\"53.6\"]},{\"cells\":[\"pi-0.5\",\"78.4\",\"73.6\",\"80.8\",\"96.2\",\"94.1\",\"89.0\",\"84.5\",\"84.4\"]},{\"cells\":[\"StarVLA\",\"52.5\",\"49.8\",\"88.5\",\"95.7\",\"95.7\",\"73.0\",\"76.9\",\"74.1\"]},{\"cells\":[\"Abot-M0\",\"60.4\",\"67.9\",\"86.4\",\"96.2\",\"91.6\",\"86.4\",\"82.6\",\"80.5\"]},{\"cells\":[\"Cosmos-Policy\",\"75.8\",\"63.3\",\"81.7\",\"96.5\",\"88.9\",\"92.7\",\"82.2\",\"82.2\"]},{\"cells\":[\"InternVLA-A1.5 (Ours)\",\"83.1\",\"55.1\",\"86.9\",\"96.4\",\"98.2\",\"95.6\",\"85.2\",\"84.8\"],\"highlight\":true}]},{\"id\":\"ebench\",\"benchmark\":\"EBench\",\"metric\":\"SR / Score\",\"value\":\"35.2 / 49.5\",\"detailCaption\":\"SR is success rate and Score is continuous task progress.\",\"detailColumns\":[\"Method\",\"Val-Train SR\",\"Val-Train Score\",\"Val-Unseen SR\",\"Val-Unseen Score\",\"Test SR\",\"Test Score\"],\"detailRows\":[{\"cells\":[\"pi-0\",\"30.5\",\"42.9\",\"25.4\",\"39.3\",\"24.4\",\"38.4\"]},{\"cells\":[\"XVLA\",\"28.3\",\"42.1\",\"22.7\",\"35.9\",\"24.7\",\"37.5\"]},{\"cells\":[\"InternVLA-A1\",\"33.1\",\"44.2\",\"20.8\",\"33.8\",\"27.6\",\"40.2\"]},{\"cells\":[\"pi-0.5\",\"32.1\",\"48.1\",\"26.5\",\"42.9\",\"29.5\",\"45.6\"]},{\"cells\":[\"LingBot-VA\",\"38.3\",\"55.9\",\"26.6\",\"43.3\",\"30.9\",\"47.6\"]},{\"cells\":[\"InternVLA-A1.5 (Ours)\",\"43.1\",\"54.1\",\"32.8\",\"46.4\",\"35.2\",\"49.5\"],\"highlight\":true}]}]}],\"$L21\"]}]]}]}]\n"])</script><script>self.__next_f.push([1,"f:[\"$\",\"section\",null,{\"className\":\"border-t border-border px-5 py-18 sm:py-24 bg-muted/35\",\"id\":\"experiments\",\"children\":[\"$\",\"div\",null,{\"className\":\"mx-auto max-w-[1120px]\",\"children\":[[\"$\",\"div\",null,{\"className\":\"mb-10 max-w-4xl\",\"children\":[[\"$\",\"p\",null,{\"className\":\"section-eyebrow mb-3\",\"children\":\"Experiments\"}],[\"$\",\"h2\",null,{\"className\":\"text-balance text-3xl font-semibold sm:text-5xl\",\"children\":\"Instruction following and long-horizon chemistry tasks\"}],[\"$\",\"p\",null,{\"className\":\"mt-5 text-lg leading-8 text-muted-foreground\",\"children\":\"The draft defines a four-task real-world suite but does not yet contain finalized success-rate numbers, so this page only describes the evaluated task setup.\"}]]}],[\"$\",\"div\",null,{\"className\":\"grid gap-8\",\"children\":[[\"$\",\"figure\",null,{\"className\":\"overflow-hidden rounded-lg border border-border bg-card p-4 shadow-sm sm:p-6\",\"children\":[[\"$\",\"$L16\",null,{\"alt\":\"Four real-world tasks for InternVLA-A1.5 evaluation.\",\"className\":\"h-auto w-full rounded-md bg-white\",\"height\":366,\"priority\":\"$undefined\",\"src\":\"/internvla-a15.github.io/reference-assets/paper/realworld-tasks.png\",\"width\":1722}],[\"$\",\"figcaption\",null,{\"className\":\"px-2 pb-1 pt-3 text-sm leading-6 text-muted-foreground\",\"children\":[\"$\",\"div\",null,{\"className\":\"real-world-caption\",\"children\":[[\"$\",\"p\",null,{\"children\":\"The real-world suite includes Sort Tubes, Insert Tubes, Move Tubes, and MOF. Colored outlines mark task-relevant tubes and targets.\"}],[\"$\",\"dl\",null,{\"className\":\"real-world-caption-grid\",\"children\":[[\"$\",\"div\",\"Sort Tubes\",{\"children\":[[\"$\",\"dt\",null,{\"children\":\"Sort Tubes\"}],[\"$\",\"dd\",null,{\"children\":\"Instruction following with a specified tube and box target, including held-out arm-color bindings for compositional grounding.\"}]]}],[\"$\",\"div\",\"Insert Tubes\",{\"children\":[[\"$\",\"dt\",null,{\"children\":\"Insert Tubes\"}],[\"$\",\"dd\",null,{\"children\":\"The robot must bring a specified tube to a specified hole of a tube rack.\"}]]}],[\"$\",\"div\",\"Move Tubes\",{\"children\":[[\"$\",\"dt\",null,{\"children\":\"Move Tubes\"}],[\"$\",\"dd\",null,{\"children\":\"The robot transports a specified tube to a specified position on the opposite side of the workspace.\"}]]}],[\"$\",\"div\",\"MOF\",{\"children\":[[\"$\",\"dt\",null,{\"children\":\"MOF\"}],[\"$\",\"dd\",null,{\"children\":\"A long-horizon chemistry procedure used to probe coherent multi-step execution.\"}]]}]]}]]}]}]]}],[\"$\",\"figure\",null,{\"className\":\"overflow-hidden rounded-lg border border-border bg-card p-4 shadow-sm sm:p-6\",\"children\":[[\"$\",\"div\",null,{\"className\":\"realworld-video-showcase\",\"children\":[[\"$\",\"div\",null,{\"className\":\"realworld-video-header\",\"children\":[[\"$\",\"h3\",null,{\"children\":\"Sort Tubes\"}],[\"$\",\"p\",null,{\"className\":\"realworld-video-description\",\"children\":\"Compositional Generalization: seen rollouts cover color and target factors; OOD rollouts swap them into held-out blue-right and orange-left instructions.\"}]]}],[\"$\",\"div\",null,{\"aria-label\":\"Seen blue plus left and seen orange plus right recombine into OOD blue plus right and OOD orange plus left.\",\"className\":\"realworld-video-formula\",\"children\":[[[\"$\",\"span\",\"Seen: blue + left\",{\"children\":\"Seen: blue + left\"}],[\"$\",\"span\",\"Seen: orange + right\",{\"children\":\"Seen: orange + right\"}]],[\"$\",\"$L17\",null,{\"ref\":\"$undefined\",\"iconNode\":[[\"path\",{\"d\":\"M5 12h14\",\"key\":\"1ays0h\"}],[\"path\",{\"d\":\"m12 5 7 7-7 7\",\"key\":\"xquz4c\"}]],\"className\":\"lucide-arrow-right realworld-video-formula-icon\",\"aria-hidden\":true}],[[\"$\",\"span\",\"OOD: blue + right\",{\"children\":\"OOD: blue + right\"}],[\"$\",\"span\",\"OOD: orange + left\",{\"children\":\"OOD: orange + left\"}]]]}],[\"$\",\"div\",null,{\"className\":\"realworld-video-scroll\",\"children\":[\"$\",\"div\",null,{\"aria-label\":\"Sort Tubes seen and OOD rollout matrix\",\"className\":\"realworld-video-matrix realworld-video-matrix--2\",\"children\":[[\"$\",\"div\",\"/internvla-a15.github.io/videos/sorttube/seen-blue-left.mp4\",{\"className\":\"realworld-video-tile\",\"data-split\":\"Seen\",\"children\":[[\"$\",\"p\",null,{\"className\":\"realworld-video-tile-label\",\"children\":[[\"$\",\"strong\",null,{\"children\":[\"Seen\",\":\"]}],[\"$\",\"span\",null,{\"children\":\"Blue tube\"}],[\"$\",\"$L17\",null,{\"ref\":\"$undefined\",\"iconNode\":\"$f:props:children:props:children:1:props:children:1:props:children:0:props:children:1:props:children:1:props:iconNode\",\"className\":\"lucide-arrow-right realworld-video-label-icon\",\"aria-hidden\":true}],\"$L22\"]}],\"$L23\"]}],\"$L24\",\"$L25\",\"$L26\"]}]}]]}],null]}],\"$L27\",\"$L28\"]}]]}]}]\n"])</script><script>self.__next_f.push([1,"10:[\"$\",\"section\",null,{\"className\":\"border-t border-border px-5 py-18 sm:py-24 bg-muted/35\",\"id\":\"bibtex\",\"children\":[\"$\",\"div\",null,{\"className\":\"mx-auto max-w-[1120px]\",\"children\":[[\"$\",\"div\",null,{\"className\":\"mb-10 max-w-4xl\",\"children\":[[\"$\",\"p\",null,{\"className\":\"section-eyebrow mb-3\",\"children\":\"Citation\"}],[\"$\",\"h2\",null,{\"className\":\"text-balance text-3xl font-semibold sm:text-5xl\",\"children\":\"BibTeX\"}],null]}],[\"$\",\"pre\",null,{\"className\":\"overflow-x-auto rounded-lg border border-border bg-card p-5 text-sm leading-7 text-foreground\",\"children\":[\"$\",\"code\",null,{\"children\":\"@misc{internvla_a15,\\n title = {InternVLA-A1.5: Unifying Understanding, Latent Foresight, and Action for Compositional Generalization},\\n author = {InternVLA-A1.5 team},\\n year = {2026},\\n howpublished = {\\\\url{https://github.com/InternRobotics/InternVLA-A1.5}}\\n}\"}]}]]}]}]\n11:[\"$\",\"script\",\"script-0\",{\"src\":\"/internvla-a15.github.io/_next/static/chunks/03xdlv9wrei1r.js\",\"async\":true,\"nonce\":\"$undefined\"}]\n12:[\"$\",\"$L29\",null,{\"children\":[\"$\",\"$2a\",null,{\"name\":\"Next.MetadataOutlet\",\"children\":\"$@2b\"}]}]\n13:[\"$\",\"$1\",\"h\",{\"children\":[null,[\"$\",\"$L2c\",null,{\"children\":\"$L2d\"}],[\"$\",\"div\",null,{\"hidden\":true,\"children\":[\"$\",\"$L2e\",null,{\"children\":[\"$\",\"$2a\",null,{\"name\":\"Next.Metadata\",\"children\":\"$L2f\"}]}]}],null]}]\n15:[\"$\",\"link\",\"0\",{\"rel\":\"stylesheet\",\"href\":\"/internvla-a15.github.io/_next/static/chunks/025_jc0.9ca-..css\",\"precedence\":\"next\",\"crossOrigin\":\"$undefined\",\"nonce\":\"$undefined\"}]\n"])</script><script>self.__next_f.push([1,"18:[\"$\",\"$L17\",null,{\"ref\":\"$undefined\",\"iconNode\":\"$8:props:children:0:props:children:0:props:iconNode\",\"className\":\"lucide-file-text\",\"aria-hidden\":\"true\"}]\n1a:[\"$\",\"td\",\"Box QA-0\",{\"children\":\"Box QA\"}]\n1b:[\"$\",\"td\",\"879K-1\",{\"children\":\"879K\"}]\n1c:[\"$\",\"td\",\"Referring detection-2\",{\"children\":\"Referring detection\"}]\n1d:[\"$\",\"tr\",\"Point QA-2\",{\"className\":\"$undefined\",\"children\":[[\"$\",\"td\",\"Point QA-0\",{\"children\":\"Point QA\"}],[\"$\",\"td\",\"832K-1\",{\"children\":\"832K\"}],[\"$\",\"td\",\"Free-space and object-point localization-2\",{\"children\":\"Free-space and object-point localization\"}]]}]\n1e:[\"$\",\"tr\",\"Trajectory QA-3\",{\"className\":\"$undefined\",\"children\":[[\"$\",\"td\",\"Trajectory QA-0\",{\"children\":\"Trajectory QA\"}],[\"$\",\"td\",\"684K-1\",{\"children\":\"684K\"}],[\"$\",\"td\",\"End-effector waypoint prediction-2\",{\"children\":\"End-effector waypoint prediction\"}]]}]\n1f:[\"$\",\"figcaption\",null,{\"className\":\"px-2 pb-1 pt-3 text-sm leading-6 text-muted-foreground\",\"children\":\"Multimodal co-training data from InternVLA-M1.\"}]\n21:[\"$\",\"figcaption\",null,{\"className\":\"px-2 pb-1 pt-3 text-sm leading-6 text-muted-foreground\",\"children\":\"Click a benchmark row to view the full table.\"}]\n22:[\"$\",\"span\",null,{\"children\":\"left box\"}]\n23:[\"$\",\"video\",null,{\"aria-label\":\"Seen rollout: Blue tube to left box\",\"autoPlay\":true,\"className\":\"realworld-video\",\"height\":720,\"loop\":true,\"muted\":true,\"playsInline\":true,\"poster\":\"/internvla-a15.github.io/videos/sorttube/seen-blue-left.jpg\",\"preload\":\"metadata\",\"width\":1280,\"children\":[[\"$\",\"source\",null,{\"src\":\"/internvla-a15.github.io/videos/sorttube/seen-blue-left.mp4\",\"type\":\"video/mp4\"}],\"Your browser does not support this video.\",\" \",[\"$\",\"a\",null,{\"href\":\"/internvla-a15.github.io/videos/sorttube/seen-blue-left.mp4\",\"children\":[\"Download this \",\"Sort Tubes\",\" rollout.\"]}]]}]\n24:[\"$\",\"div\",\"/internvla-a15.github.io/videos/sorttube/ood-blue-right.mp4\",{\"className\":\"realworld-video-tile\",\"data-split\":\"OOD\",\"children\":[[\"$\",\"p\",null,{\"className\":\"realworld-video-tile-label\",\"children\":[[\"$\",\"strong\",null,{\"children\":[\"OOD\",\":\"]}],[\"$\",\"span\",null,{\"children\":\"Blue tube\"}],[\"$\",\"$L17\",null,{\"ref\":\"$undefined\",\"iconNode\":\"$f:props:children:props:children:1:props:children:1:props:children:0:props:children:1:props:children:1:props:iconNode\",\"className\":\"lucide-arrow-right realworld-video-label-icon\",\"aria-hidden\":true}],[\"$\",\"span\",null,{\"children\":\"right box\"}]]}],[\"$\",\"video\",null,{\"aria-label\":\"OOD rollout: Blue tube to right box\",\"autoPlay\":true,\"className\":\"realworld-video\",\"height\":720,\"loop\":true,\"muted\":true,\"playsInline\":true,\"poster\":\"/internvla-a15.github.io/videos/sorttube/ood-blue-right.jpg\",\"preload\":\"metadata\",\"width\":1280,\"children\":[[\"$\",\"source\",null,{\"src\":\"/internvla-a15.github.io/videos/sorttube/ood-blue-right.mp4\",\"type\":\"video/mp4\"}],\"Your browser does not support this video.\",\" \",[\"$\",\"a\",null,{\"href\":\"/internvla-a15.github.io/videos/sorttube/ood-blue-right.mp4\",\"children\":[\"Download this \",\"Sort Tubes\",\" rollout.\"]}]]}]]}]\n25:[\"$\",\"div\",\"/internvla-a15.github.io/videos/sorttube/ood-orange-left.mp4\",{\"className\":\"realworld-video-tile\",\"data-split\":\"OOD\",\"children\":[[\"$\",\"p\",null,{\"className\":\"realworld-video-tile-label\",\"children\":[[\"$\",\"strong\",null,{\"children\":[\"OOD\",\":\"]}],[\"$\",\"span\",null,{\"children\":\"Orange tube\"}],[\"$\",\"$L17\",null,{\"ref\":\"$undefined\",\"iconNode\":\"$f:props:children:props:children:1:props:children:1:props:children:0:props:children:1:props:children:1:props:iconNode\",\"className\":\"lucide-arrow-right realworld-video-label-icon\",\"aria-hidden\":true}],[\"$\",\"span\",null,{\"children\":\"left box\"}]]}],[\"$\",\"video\",null,{\"aria-label\":\"OOD rollout: Orange tube to left box\",\"autoPlay\":true,\"className\":\"realworld-video\",\"height\":720,\"loop\":true,\"muted\":true,\"playsInline\":true,\"poster\":\"/internvla-a15.github.io/videos/sorttube/ood-orange-left.jpg\",\"preload\":\"metadata\",\"width\":1280,\"children\":[[\"$\",\"source\",null,{\"src\":\"/internvla-a15.github.io/videos/sorttube/ood-orange-left.mp4\",\"type\":\"video/mp4\"}],\"Your browser does not support this video.\",\" \",[\"$\",\"a\",null,{\"href\":\"/internvla-a15.github.io/video"])</script><script>self.__next_f.push([1,"s/sorttube/ood-orange-left.mp4\",\"children\":[\"Download this \",\"Sort Tubes\",\" rollout.\"]}]]}]]}]\n26:[\"$\",\"div\",\"/internvla-a15.github.io/videos/sorttube/seen-orange-right.mp4\",{\"className\":\"realworld-video-tile\",\"data-split\":\"Seen\",\"children\":[[\"$\",\"p\",null,{\"className\":\"realworld-video-tile-label\",\"children\":[[\"$\",\"strong\",null,{\"children\":[\"Seen\",\":\"]}],[\"$\",\"span\",null,{\"children\":\"Orange tube\"}],[\"$\",\"$L17\",null,{\"ref\":\"$undefined\",\"iconNode\":\"$f:props:children:props:children:1:props:children:1:props:children:0:props:children:1:props:children:1:props:iconNode\",\"className\":\"lucide-arrow-right realworld-video-label-icon\",\"aria-hidden\":true}],[\"$\",\"span\",null,{\"children\":\"right box\"}]]}],[\"$\",\"video\",null,{\"aria-label\":\"Seen rollout: Orange tube to right box\",\"autoPlay\":true,\"className\":\"realworld-video\",\"height\":720,\"loop\":true,\"muted\":true,\"playsInline\":true,\"poster\":\"/internvla-a15.github.io/videos/sorttube/seen-orange-right.jpg\",\"preload\":\"metadata\",\"width\":1280,\"children\":[[\"$\",\"source\",null,{\"src\":\"/internvla-a15.github.io/videos/sorttube/seen-orange-right.mp4\",\"type\":\"video/mp4\"}],\"Your browser does not support this video.\",\" \",[\"$\",\"a\",null,{\"href\":\"/internvla-a15.github.io/videos/sorttube/seen-orange-right.mp4\",\"children\":[\"Download this \",\"Sort Tubes\",\" rollout.\"]}]]}]]}]\n"])</script><script>self.__next_f.push([1,"27:[\"$\",\"figure\",null,{\"className\":\"overflow-hidden rounded-lg border border-border bg-card p-4 shadow-sm sm:p-6\",\"children\":[[\"$\",\"div\",null,{\"className\":\"realworld-video-showcase\",\"children\":[[\"$\",\"div\",null,{\"className\":\"realworld-video-header\",\"children\":[[\"$\",\"h3\",null,{\"children\":\"Insert Tubes\"}],[\"$\",\"p\",null,{\"className\":\"realworld-video-description\",\"children\":\"Compositional Generalization: the policy binds tube color to hole index and transfers that binding to held-out color-hole pairings.\"}]]}],null,[\"$\",\"div\",null,{\"className\":\"realworld-video-scroll\",\"children\":[\"$\",\"div\",null,{\"aria-label\":\"Insert Tubes row-major rollout matrix\",\"className\":\"realworld-video-matrix realworld-video-matrix--4\",\"children\":[[\"$\",\"div\",\"/internvla-a15.github.io/videos/inserttube/seen-b1.mp4\",{\"className\":\"realworld-video-tile\",\"data-split\":\"Seen\",\"children\":[[\"$\",\"p\",null,{\"className\":\"realworld-video-tile-label\",\"children\":[[\"$\",\"strong\",null,{\"children\":[\"Seen\",\":\"]}],[\"$\",\"span\",null,{\"children\":\"Blue\"}],[\"$\",\"$L17\",null,{\"ref\":\"$undefined\",\"iconNode\":\"$f:props:children:props:children:1:props:children:1:props:children:0:props:children:1:props:children:1:props:iconNode\",\"className\":\"lucide-arrow-right realworld-video-label-icon\",\"aria-hidden\":true}],[\"$\",\"span\",null,{\"children\":\"Hole 1\"}]]}],[\"$\",\"video\",null,{\"aria-label\":\"Seen rollout: Blue tube to Hole 1\",\"autoPlay\":true,\"className\":\"realworld-video\",\"height\":720,\"loop\":true,\"muted\":true,\"playsInline\":true,\"poster\":\"/internvla-a15.github.io/videos/inserttube/seen-b1.jpg\",\"preload\":\"metadata\",\"width\":1280,\"children\":[[\"$\",\"source\",null,{\"src\":\"/internvla-a15.github.io/videos/inserttube/seen-b1.mp4\",\"type\":\"video/mp4\"}],\"Your browser does not support this video.\",\" \",[\"$\",\"a\",null,{\"href\":\"/internvla-a15.github.io/videos/inserttube/seen-b1.mp4\",\"children\":[\"Download this \",\"Insert Tubes\",\" rollout.\"]}]]}]]}],[\"$\",\"div\",\"/internvla-a15.github.io/videos/inserttube/ood-b2.mp4\",{\"className\":\"realworld-video-tile\",\"data-split\":\"OOD\",\"children\":[[\"$\",\"p\",null,{\"className\":\"realworld-video-tile-label\",\"children\":[[\"$\",\"strong\",null,{\"children\":[\"OOD\",\":\"]}],[\"$\",\"span\",null,{\"children\":\"Blue\"}],[\"$\",\"$L17\",null,{\"ref\":\"$undefined\",\"iconNode\":\"$f:props:children:props:children:1:props:children:1:props:children:0:props:children:1:props:children:1:props:iconNode\",\"className\":\"lucide-arrow-right realworld-video-label-icon\",\"aria-hidden\":true}],[\"$\",\"span\",null,{\"children\":\"Hole 2\"}]]}],[\"$\",\"video\",null,{\"aria-label\":\"OOD rollout: Blue tube to Hole 2\",\"autoPlay\":true,\"className\":\"realworld-video\",\"height\":720,\"loop\":true,\"muted\":true,\"playsInline\":true,\"poster\":\"/internvla-a15.github.io/videos/inserttube/ood-b2.jpg\",\"preload\":\"metadata\",\"width\":1280,\"children\":[[\"$\",\"source\",null,{\"src\":\"/internvla-a15.github.io/videos/inserttube/ood-b2.mp4\",\"type\":\"video/mp4\"}],\"Your browser does not support this video.\",\" \",[\"$\",\"a\",null,{\"href\":\"/internvla-a15.github.io/videos/inserttube/ood-b2.mp4\",\"children\":[\"Download this \",\"Insert Tubes\",\" rollout.\"]}]]}]]}],[\"$\",\"div\",\"/internvla-a15.github.io/videos/inserttube/seen-b3.mp4\",{\"className\":\"realworld-video-tile\",\"data-split\":\"Seen\",\"children\":[[\"$\",\"p\",null,{\"className\":\"realworld-video-tile-label\",\"children\":[[\"$\",\"strong\",null,{\"children\":[\"Seen\",\":\"]}],[\"$\",\"span\",null,{\"children\":\"Blue\"}],[\"$\",\"$L17\",null,{\"ref\":\"$undefined\",\"iconNode\":\"$f:props:children:props:children:1:props:children:1:props:children:0:props:children:1:props:children:1:props:iconNode\",\"className\":\"lucide-arrow-right realworld-video-label-icon\",\"aria-hidden\":true}],[\"$\",\"span\",null,{\"children\":\"Hole 3\"}]]}],[\"$\",\"video\",null,{\"aria-label\":\"Seen rollout: Blue tube to Hole 3\",\"autoPlay\":true,\"className\":\"realworld-video\",\"height\":720,\"loop\":true,\"muted\":true,\"playsInline\":true,\"poster\":\"/internvla-a15.github.io/videos/inserttube/seen-b3.jpg\",\"preload\":\"metadata\",\"width\":1280,\"children\":[[\"$\",\"source\",null,{\"src\":\"/internvla-a15.github.io/videos/inserttube/seen-b3.mp4\",\"type\":\"video/mp4\"}],\"Your browser does not support this video.\",\" \",[\"$\",\"a\",null,{\"href\":\"/internvla-a15.github.io/videos/inserttube/seen-b3.mp4\",\"children\":[\"Download this \",\"Insert Tubes\",\" rollout.\"]}]]}]]}],[\"$\",\"div\",\"/internvla-a15.github.io/videos/inserttube/seen-b4.mp4\",{\"className\":\"realworld-video-tile\",\"data-split\":\"Seen\",\"children\":[[\"$\",\"p\",null,{\"className\":\"realworld-video-tile-label\",\"children\":[[\"$\",\"strong\",null,{\"children\":[\"Seen\",\":\"]}],[\"$\",\"span\",null,{\"children\":\"Blue\"}],[\"$\",\"$L17\",null,{\"ref\":\"$undefined\",\"iconNode\":\"$f:props:children:props:children:1:props:children:1:props:children:0:props:children:1:props:children:1:props:iconNode\",\"className\":\"lucide-arrow-right realworld-video-label-icon\",\"aria-hidden\":true}],[\"$\",\"span\",null,{\"children\":\"Hole 4\"}]]}],[\"$\",\"video\",null,{\"aria-label\":\"Seen rollout: Blue tube to Hole 4\",\"autoPlay\":true,\"className\":\"realworld-video\",\"height\":720,\"loop\":true,\"muted\":true,\"playsInline\":true,\"poster\":\"/internvla-a15.github.io/videos/inserttube/seen-b4.jpg\",\"preload\":\"metadata\",\"width\":1280,\"children\":[\"$L30\",\"Your browser does not support this video.\",\" \",\"$L31\"]}]]}],\"$L32\",\"$L33\",\"$L34\",\"$L35\"]}]}]]}],null]}]\n"])</script><script>self.__next_f.push([1,"28:[\"$\",\"figure\",null,{\"className\":\"overflow-hidden rounded-lg border border-border bg-card p-4 shadow-sm sm:p-6\",\"children\":[[\"$\",\"div\",null,{\"className\":\"realworld-video-showcase\",\"children\":[[\"$\",\"div\",null,{\"className\":\"realworld-video-header\",\"children\":[[\"$\",\"h3\",null,{\"children\":\"Move Tubes\"}],[\"$\",\"p\",null,{\"className\":\"realworld-video-description\",\"children\":\"Compositional Generalization: the model reuses color and target-hole factors across a different manipulation primitive and OOD pairings.\"}]]}],null,[\"$\",\"div\",null,{\"className\":\"realworld-video-scroll\",\"children\":[\"$\",\"div\",null,{\"aria-label\":\"Move Tubes row-major rollout matrix\",\"className\":\"realworld-video-matrix realworld-video-matrix--4\",\"children\":[[\"$\",\"div\",\"/internvla-a15.github.io/videos/movetube/ood-b1.mp4\",{\"className\":\"realworld-video-tile\",\"data-split\":\"OOD\",\"children\":[[\"$\",\"p\",null,{\"className\":\"realworld-video-tile-label\",\"children\":[[\"$\",\"strong\",null,{\"children\":[\"OOD\",\":\"]}],[\"$\",\"span\",null,{\"children\":\"Blue\"}],[\"$\",\"$L17\",null,{\"ref\":\"$undefined\",\"iconNode\":\"$f:props:children:props:children:1:props:children:1:props:children:0:props:children:1:props:children:1:props:iconNode\",\"className\":\"lucide-arrow-right realworld-video-label-icon\",\"aria-hidden\":true}],[\"$\",\"span\",null,{\"children\":\"Hole 1\"}]]}],[\"$\",\"video\",null,{\"aria-label\":\"OOD rollout: Blue tube to Hole 1\",\"autoPlay\":true,\"className\":\"realworld-video\",\"height\":720,\"loop\":true,\"muted\":true,\"playsInline\":true,\"poster\":\"/internvla-a15.github.io/videos/movetube/ood-b1.jpg\",\"preload\":\"metadata\",\"width\":1280,\"children\":[[\"$\",\"source\",null,{\"src\":\"/internvla-a15.github.io/videos/movetube/ood-b1.mp4\",\"type\":\"video/mp4\"}],\"Your browser does not support this video.\",\" \",[\"$\",\"a\",null,{\"href\":\"/internvla-a15.github.io/videos/movetube/ood-b1.mp4\",\"children\":[\"Download this \",\"Move Tubes\",\" rollout.\"]}]]}]]}],[\"$\",\"div\",\"/internvla-a15.github.io/videos/movetube/seen-b2.mp4\",{\"className\":\"realworld-video-tile\",\"data-split\":\"Seen\",\"children\":[[\"$\",\"p\",null,{\"className\":\"realworld-video-tile-label\",\"children\":[[\"$\",\"strong\",null,{\"children\":[\"Seen\",\":\"]}],[\"$\",\"span\",null,{\"children\":\"Blue\"}],[\"$\",\"$L17\",null,{\"ref\":\"$undefined\",\"iconNode\":\"$f:props:children:props:children:1:props:children:1:props:children:0:props:children:1:props:children:1:props:iconNode\",\"className\":\"lucide-arrow-right realworld-video-label-icon\",\"aria-hidden\":true}],[\"$\",\"span\",null,{\"children\":\"Hole 2\"}]]}],[\"$\",\"video\",null,{\"aria-label\":\"Seen rollout: Blue tube to Hole 2\",\"autoPlay\":true,\"className\":\"realworld-video\",\"height\":720,\"loop\":true,\"muted\":true,\"playsInline\":true,\"poster\":\"/internvla-a15.github.io/videos/movetube/seen-b2.jpg\",\"preload\":\"metadata\",\"width\":1280,\"children\":[[\"$\",\"source\",null,{\"src\":\"/internvla-a15.github.io/videos/movetube/seen-b2.mp4\",\"type\":\"video/mp4\"}],\"Your browser does not support this video.\",\" \",[\"$\",\"a\",null,{\"href\":\"/internvla-a15.github.io/videos/movetube/seen-b2.mp4\",\"children\":[\"Download this \",\"Move Tubes\",\" rollout.\"]}]]}]]}],[\"$\",\"div\",\"/internvla-a15.github.io/videos/movetube/ood-b3.mp4\",{\"className\":\"realworld-video-tile\",\"data-split\":\"OOD\",\"children\":[[\"$\",\"p\",null,{\"className\":\"realworld-video-tile-label\",\"children\":[[\"$\",\"strong\",null,{\"children\":[\"OOD\",\":\"]}],[\"$\",\"span\",null,{\"children\":\"Blue\"}],[\"$\",\"$L17\",null,{\"ref\":\"$undefined\",\"iconNode\":\"$f:props:children:props:children:1:props:children:1:props:children:0:props:children:1:props:children:1:props:iconNode\",\"className\":\"lucide-arrow-right realworld-video-label-icon\",\"aria-hidden\":true}],[\"$\",\"span\",null,{\"children\":\"Hole 3\"}]]}],[\"$\",\"video\",null,{\"aria-label\":\"OOD rollout: Blue tube to Hole 3\",\"autoPlay\":true,\"className\":\"realworld-video\",\"height\":720,\"loop\":true,\"muted\":true,\"playsInline\":true,\"poster\":\"/internvla-a15.github.io/videos/movetube/ood-b3.jpg\",\"preload\":\"metadata\",\"width\":1280,\"children\":[[\"$\",\"source\",null,{\"src\":\"/internvla-a15.github.io/videos/movetube/ood-b3.mp4\",\"type\":\"video/mp4\"}],\"Your browser does not support this video.\",\" \",[\"$\",\"a\",null,{\"href\":\"/internvla-a15.github.io/videos/movetube/ood-b3.mp4\",\"children\":[\"Download this \",\"Move Tubes\",\" rollout.\"]}]]}]]}],[\"$\",\"div\",\"/internvla-a15.github.io/videos/movetube/seen-b4.mp4\",{\"className\":\"realworld-video-tile\",\"data-split\":\"Seen\",\"children\":[[\"$\",\"p\",null,{\"className\":\"realworld-video-tile-label\",\"children\":[[\"$\",\"strong\",null,{\"children\":[\"Seen\",\":\"]}],[\"$\",\"span\",null,{\"children\":\"Blue\"}],[\"$\",\"$L17\",null,{\"ref\":\"$undefined\",\"iconNode\":\"$f:props:children:props:children:1:props:children:1:props:children:0:props:children:1:props:children:1:props:iconNode\",\"className\":\"lucide-arrow-right realworld-video-label-icon\",\"aria-hidden\":true}],[\"$\",\"span\",null,{\"children\":\"Hole 4\"}]]}],[\"$\",\"video\",null,{\"aria-label\":\"Seen rollout: Blue tube to Hole 4\",\"autoPlay\":true,\"className\":\"realworld-video\",\"height\":720,\"loop\":true,\"muted\":true,\"playsInline\":true,\"poster\":\"/internvla-a15.github.io/videos/movetube/seen-b4.jpg\",\"preload\":\"metadata\",\"width\":1280,\"children\":[\"$L36\",\"Your browser does not support this video.\",\" \",\"$L37\"]}]]}],\"$L38\",\"$L39\",\"$L3a\",\"$L3b\"]}]}]]}],null]}]\n"])</script><script>self.__next_f.push([1,"30:[\"$\",\"source\",null,{\"src\":\"/internvla-a15.github.io/videos/inserttube/seen-b4.mp4\",\"type\":\"video/mp4\"}]\n31:[\"$\",\"a\",null,{\"href\":\"/internvla-a15.github.io/videos/inserttube/seen-b4.mp4\",\"children\":[\"Download this \",\"Insert Tubes\",\" rollout.\"]}]\n32:[\"$\",\"div\",\"/internvla-a15.github.io/videos/inserttube/seen-o1.mp4\",{\"className\":\"realworld-video-tile\",\"data-split\":\"Seen\",\"children\":[[\"$\",\"p\",null,{\"className\":\"realworld-video-tile-label\",\"children\":[[\"$\",\"strong\",null,{\"children\":[\"Seen\",\":\"]}],[\"$\",\"span\",null,{\"children\":\"Orange\"}],[\"$\",\"$L17\",null,{\"ref\":\"$undefined\",\"iconNode\":\"$f:props:children:props:children:1:props:children:1:props:children:0:props:children:1:props:children:1:props:iconNode\",\"className\":\"lucide-arrow-right realworld-video-label-icon\",\"aria-hidden\":true}],[\"$\",\"span\",null,{\"children\":\"Hole 1\"}]]}],[\"$\",\"video\",null,{\"aria-label\":\"Seen rollout: Orange tube to Hole 1\",\"autoPlay\":true,\"className\":\"realworld-video\",\"height\":720,\"loop\":true,\"muted\":true,\"playsInline\":true,\"poster\":\"/internvla-a15.github.io/videos/inserttube/seen-o1.jpg\",\"preload\":\"metadata\",\"width\":1280,\"children\":[[\"$\",\"source\",null,{\"src\":\"/internvla-a15.github.io/videos/inserttube/seen-o1.mp4\",\"type\":\"video/mp4\"}],\"Your browser does not support this video.\",\" \",[\"$\",\"a\",null,{\"href\":\"/internvla-a15.github.io/videos/inserttube/seen-o1.mp4\",\"children\":[\"Download this \",\"Insert Tubes\",\" rollout.\"]}]]}]]}]\n33:[\"$\",\"div\",\"/internvla-a15.github.io/videos/inserttube/seen-o2.mp4\",{\"className\":\"realworld-video-tile\",\"data-split\":\"Seen\",\"children\":[[\"$\",\"p\",null,{\"className\":\"realworld-video-tile-label\",\"children\":[[\"$\",\"strong\",null,{\"children\":[\"Seen\",\":\"]}],[\"$\",\"span\",null,{\"children\":\"Orange\"}],[\"$\",\"$L17\",null,{\"ref\":\"$undefined\",\"iconNode\":\"$f:props:children:props:children:1:props:children:1:props:children:0:props:children:1:props:children:1:props:iconNode\",\"className\":\"lucide-arrow-right realworld-video-label-icon\",\"aria-hidden\":true}],[\"$\",\"span\",null,{\"children\":\"Hole 2\"}]]}],[\"$\",\"video\",null,{\"aria-label\":\"Seen rollout: Orange tube to Hole 2\",\"autoPlay\":true,\"className\":\"realworld-video\",\"height\":720,\"loop\":true,\"muted\":true,\"playsInline\":true,\"poster\":\"/internvla-a15.github.io/videos/inserttube/seen-o2.jpg\",\"preload\":\"metadata\",\"width\":1280,\"children\":[[\"$\",\"source\",null,{\"src\":\"/internvla-a15.github.io/videos/inserttube/seen-o2.mp4\",\"type\":\"video/mp4\"}],\"Your browser does not support this video.\",\" \",[\"$\",\"a\",null,{\"href\":\"/internvla-a15.github.io/videos/inserttube/seen-o2.mp4\",\"children\":[\"Download this \",\"Insert Tubes\",\" rollout.\"]}]]}]]}]\n34:[\"$\",\"div\",\"/internvla-a15.github.io/videos/inserttube/seen-o3.mp4\",{\"className\":\"realworld-video-tile\",\"data-split\":\"Seen\",\"children\":[[\"$\",\"p\",null,{\"className\":\"realworld-video-tile-label\",\"children\":[[\"$\",\"strong\",null,{\"children\":[\"Seen\",\":\"]}],[\"$\",\"span\",null,{\"children\":\"Orange\"}],[\"$\",\"$L17\",null,{\"ref\":\"$undefined\",\"iconNode\":\"$f:props:children:props:children:1:props:children:1:props:children:0:props:children:1:props:children:1:props:iconNode\",\"className\":\"lucide-arrow-right realworld-video-label-icon\",\"aria-hidden\":true}],[\"$\",\"span\",null,{\"children\":\"Hole 3\"}]]}],[\"$\",\"video\",null,{\"aria-label\":\"Seen rollout: Orange tube to Hole 3\",\"autoPlay\":true,\"className\":\"realworld-video\",\"height\":720,\"loop\":true,\"muted\":true,\"playsInline\":true,\"poster\":\"/internvla-a15.github.io/videos/inserttube/seen-o3.jpg\",\"preload\":\"metadata\",\"width\":1280,\"children\":[[\"$\",\"source\",null,{\"src\":\"/internvla-a15.github.io/videos/inserttube/seen-o3.mp4\",\"type\":\"video/mp4\"}],\"Your browser does not support this video.\",\" \",[\"$\",\"a\",null,{\"href\":\"/internvla-a15.github.io/videos/inserttube/seen-o3.mp4\",\"children\":[\"Download this \",\"Insert Tubes\",\" rollout.\"]}]]}]]}]\n35:[\"$\",\"div\",\"/internvla-a15.github.io/videos/inserttube/ood-o4.mp4\",{\"className\":\"realworld-video-tile\",\"data-split\":\"OOD\",\"children\":[[\"$\",\"p\",null,{\"className\":\"realworld-video-tile-label\",\"children\":[[\"$\",\"strong\",null,{\"children\":[\"OOD\",\":\"]}],[\"$\",\"span\",null,{\"children\":\"Orange\"}],[\"$\",\"$L17\",null,{\"ref\":\"$undefined\",\"iconNode\":\"$f:props:ch"])</script><script>self.__next_f.push([1,"ildren:props:children:1:props:children:1:props:children:0:props:children:1:props:children:1:props:iconNode\",\"className\":\"lucide-arrow-right realworld-video-label-icon\",\"aria-hidden\":true}],[\"$\",\"span\",null,{\"children\":\"Hole 4\"}]]}],[\"$\",\"video\",null,{\"aria-label\":\"OOD rollout: Orange tube to Hole 4\",\"autoPlay\":true,\"className\":\"realworld-video\",\"height\":720,\"loop\":true,\"muted\":true,\"playsInline\":true,\"poster\":\"/internvla-a15.github.io/videos/inserttube/ood-o4.jpg\",\"preload\":\"metadata\",\"width\":1280,\"children\":[[\"$\",\"source\",null,{\"src\":\"/internvla-a15.github.io/videos/inserttube/ood-o4.mp4\",\"type\":\"video/mp4\"}],\"Your browser does not support this video.\",\" \",[\"$\",\"a\",null,{\"href\":\"/internvla-a15.github.io/videos/inserttube/ood-o4.mp4\",\"children\":[\"Download this \",\"Insert Tubes\",\" rollout.\"]}]]}]]}]\n36:[\"$\",\"source\",null,{\"src\":\"/internvla-a15.github.io/videos/movetube/seen-b4.mp4\",\"type\":\"video/mp4\"}]\n37:[\"$\",\"a\",null,{\"href\":\"/internvla-a15.github.io/videos/movetube/seen-b4.mp4\",\"children\":[\"Download this \",\"Move Tubes\",\" rollout.\"]}]\n38:[\"$\",\"div\",\"/internvla-a15.github.io/videos/movetube/seen-o1.mp4\",{\"className\":\"realworld-video-tile\",\"data-split\":\"Seen\",\"children\":[[\"$\",\"p\",null,{\"className\":\"realworld-video-tile-label\",\"children\":[[\"$\",\"strong\",null,{\"children\":[\"Seen\",\":\"]}],[\"$\",\"span\",null,{\"children\":\"Orange\"}],[\"$\",\"$L17\",null,{\"ref\":\"$undefined\",\"iconNode\":\"$f:props:children:props:children:1:props:children:1:props:children:0:props:children:1:props:children:1:props:iconNode\",\"className\":\"lucide-arrow-right realworld-video-label-icon\",\"aria-hidden\":true}],[\"$\",\"span\",null,{\"children\":\"Hole 1\"}]]}],[\"$\",\"video\",null,{\"aria-label\":\"Seen rollout: Orange tube to Hole 1\",\"autoPlay\":true,\"className\":\"realworld-video\",\"height\":720,\"loop\":true,\"muted\":true,\"playsInline\":true,\"poster\":\"/internvla-a15.github.io/videos/movetube/seen-o1.jpg\",\"preload\":\"metadata\",\"width\":1280,\"children\":[[\"$\",\"source\",null,{\"src\":\"/internvla-a15.github.io/videos/movetube/seen-o1.mp4\",\"type\":\"video/mp4\"}],\"Your browser does not support this video.\",\" \",[\"$\",\"a\",null,{\"href\":\"/internvla-a15.github.io/videos/movetube/seen-o1.mp4\",\"children\":[\"Download this \",\"Move Tubes\",\" rollout.\"]}]]}]]}]\n39:[\"$\",\"div\",\"/internvla-a15.github.io/videos/movetube/ood-o2.mp4\",{\"className\":\"realworld-video-tile\",\"data-split\":\"OOD\",\"children\":[[\"$\",\"p\",null,{\"className\":\"realworld-video-tile-label\",\"children\":[[\"$\",\"strong\",null,{\"children\":[\"OOD\",\":\"]}],[\"$\",\"span\",null,{\"children\":\"Orange\"}],[\"$\",\"$L17\",null,{\"ref\":\"$undefined\",\"iconNode\":\"$f:props:children:props:children:1:props:children:1:props:children:0:props:children:1:props:children:1:props:iconNode\",\"className\":\"lucide-arrow-right realworld-video-label-icon\",\"aria-hidden\":true}],[\"$\",\"span\",null,{\"children\":\"Hole 2\"}]]}],[\"$\",\"video\",null,{\"aria-label\":\"OOD rollout: Orange tube to Hole 2\",\"autoPlay\":true,\"className\":\"realworld-video\",\"height\":720,\"loop\":true,\"muted\":true,\"playsInline\":true,\"poster\":\"/internvla-a15.github.io/videos/movetube/ood-o2.jpg\",\"preload\":\"metadata\",\"width\":1280,\"children\":[[\"$\",\"source\",null,{\"src\":\"/internvla-a15.github.io/videos/movetube/ood-o2.mp4\",\"type\":\"video/mp4\"}],\"Your browser does not support this video.\",\" \",[\"$\",\"a\",null,{\"href\":\"/internvla-a15.github.io/videos/movetube/ood-o2.mp4\",\"children\":[\"Download this \",\"Move Tubes\",\" rollout.\"]}]]}]]}]\n3a:[\"$\",\"div\",\"/internvla-a15.github.io/videos/movetube/seen-o3.mp4\",{\"className\":\"realworld-video-tile\",\"data-split\":\"Seen\",\"children\":[[\"$\",\"p\",null,{\"className\":\"realworld-video-tile-label\",\"children\":[[\"$\",\"strong\",null,{\"children\":[\"Seen\",\":\"]}],[\"$\",\"span\",null,{\"children\":\"Orange\"}],[\"$\",\"$L17\",null,{\"ref\":\"$undefined\",\"iconNode\":\"$f:props:children:props:children:1:props:children:1:props:children:0:props:children:1:props:children:1:props:iconNode\",\"className\":\"lucide-arrow-right realworld-video-label-icon\",\"aria-hidden\":true}],[\"$\",\"span\",null,{\"children\":\"Hole 3\"}]]}],[\"$\",\"video\",null,{\"aria-label\":\"Seen rollout: Orange tube to Hole 3\",\"autoPlay\":true,\"className\":\"realworld-video\",\"height\":720,\"loop\":true,\"muted\":true,\"pl"])</script><script>self.__next_f.push([1,"aysInline\":true,\"poster\":\"/internvla-a15.github.io/videos/movetube/seen-o3.jpg\",\"preload\":\"metadata\",\"width\":1280,\"children\":[[\"$\",\"source\",null,{\"src\":\"/internvla-a15.github.io/videos/movetube/seen-o3.mp4\",\"type\":\"video/mp4\"}],\"Your browser does not support this video.\",\" \",[\"$\",\"a\",null,{\"href\":\"/internvla-a15.github.io/videos/movetube/seen-o3.mp4\",\"children\":[\"Download this \",\"Move Tubes\",\" rollout.\"]}]]}]]}]\n3b:[\"$\",\"div\",\"/internvla-a15.github.io/videos/movetube/ood-o4.mp4\",{\"className\":\"realworld-video-tile\",\"data-split\":\"OOD\",\"children\":[[\"$\",\"p\",null,{\"className\":\"realworld-video-tile-label\",\"children\":[[\"$\",\"strong\",null,{\"children\":[\"OOD\",\":\"]}],[\"$\",\"span\",null,{\"children\":\"Orange\"}],[\"$\",\"$L17\",null,{\"ref\":\"$undefined\",\"iconNode\":\"$f:props:children:props:children:1:props:children:1:props:children:0:props:children:1:props:children:1:props:iconNode\",\"className\":\"lucide-arrow-right realworld-video-label-icon\",\"aria-hidden\":true}],[\"$\",\"span\",null,{\"children\":\"Hole 4\"}]]}],[\"$\",\"video\",null,{\"aria-label\":\"OOD rollout: Orange tube to Hole 4\",\"autoPlay\":true,\"className\":\"realworld-video\",\"height\":720,\"loop\":true,\"muted\":true,\"playsInline\":true,\"poster\":\"/internvla-a15.github.io/videos/movetube/ood-o4.jpg\",\"preload\":\"metadata\",\"width\":1280,\"children\":[[\"$\",\"source\",null,{\"src\":\"/internvla-a15.github.io/videos/movetube/ood-o4.mp4\",\"type\":\"video/mp4\"}],\"Your browser does not support this video.\",\" \",[\"$\",\"a\",null,{\"href\":\"/internvla-a15.github.io/videos/movetube/ood-o4.mp4\",\"children\":[\"Download this \",\"Move Tubes\",\" rollout.\"]}]]}]]}]\n"])</script><script>self.__next_f.push([1,"2d:[[\"$\",\"meta\",\"0\",{\"charSet\":\"utf-8\"}],[\"$\",\"meta\",\"1\",{\"name\":\"viewport\",\"content\":\"width=device-width, initial-scale=1\"}],[\"$\",\"meta\",\"2\",{\"name\":\"theme-color\",\"content\":\"#fbfbfd\",\"media\":\"(prefers-color-scheme: light)\"}],[\"$\",\"meta\",\"3\",{\"name\":\"theme-color\",\"content\":\"#0d0f17\",\"media\":\"(prefers-color-scheme: dark)\"}],[\"$\",\"meta\",\"4\",{\"name\":\"color-scheme\",\"content\":\"light dark\"}]]\n"])</script><script>self.__next_f.push([1,"2b:null\n2f:[[\"$\",\"title\",\"0\",{\"children\":\"InternVLA-A1.5\"}],[\"$\",\"meta\",\"1\",{\"name\":\"description\",\"content\":\"InternVLA-A1.5 unifies vision-language understanding, latent visual foresight, and action generation for robot policy learning.\"}],[\"$\",\"meta\",\"2\",{\"name\":\"application-name\",\"content\":\"InternVLA-A1.5\"}],[\"$\",\"meta\",\"3\",{\"name\":\"creator\",\"content\":\"InternVLA-A1.5 Team\"}],[\"$\",\"meta\",\"4\",{\"property\":\"og:title\",\"content\":\"InternVLA-A1.5\"}],[\"$\",\"meta\",\"5\",{\"property\":\"og:description\",\"content\":\"InternVLA-A1.5 unifies vision-language understanding, latent visual foresight, and action generation for robot policy learning.\"}],[\"$\",\"meta\",\"6\",{\"property\":\"og:site_name\",\"content\":\"InternVLA-A1.5\"}],[\"$\",\"meta\",\"7\",{\"property\":\"og:type\",\"content\":\"website\"}],[\"$\",\"meta\",\"8\",{\"name\":\"twitter:card\",\"content\":\"summary_large_image\"}],[\"$\",\"meta\",\"9\",{\"name\":\"twitter:title\",\"content\":\"InternVLA-A1.5\"}],[\"$\",\"meta\",\"10\",{\"name\":\"twitter:description\",\"content\":\"InternVLA-A1.5 unifies vision-language understanding, latent visual foresight, and action generation for robot policy learning.\"}]]\n"])</script></body></html>