DX-596-together-dedicated-endpoints: sync with mintlify-docs#1215

[zainhas]

Syncs together-dedicated-endpoints with togethercomputer/mintlify-docs#1215 — MOSH-3229: list Multi-LoRA supported models and target modules.

Changed docs that triggered this sync

• docs/dedicated-endpoints/lora-adapter.mdx — adds a "Supported models" section listing LoRA-enabled base models, their supported target modules, and the max loaded adapters per endpoint.

Skill changes

• Add a new LoRA-Enabled Base Models section to references/api-reference.md with a table of LoRA-enabled base models, supported LoRA target modules, and max loaded adapters per endpoint (Gemma 3 / Gemma 4, Mixtral-8x7B FP8, Llama 3.3 70B FP8, Llama 4 Scout FP8, and the Qwen 3.5 / Qwen 3.6 dense + MoE family).
• Note that Qwen 3.5 / Qwen 3.6 adapters target only the standard attention (q_proj, k_proj, v_proj, o_proj) and MLP (gate_proj, up_proj, down_proj) projections — adapters that target other layers (e.g. the vision encoder) may not load for serving.
• Add the new section to the reference's Contents ToC.
• Add a Quick Routing bullet in SKILL.md pointing to the new section for questions about target modules and max loaded adapters.

Note: #44 (Multi-LoRA Adapters add/list/remove API surface, from mintlify-docs#878) is still open against the same file. This PR adds a self-contained Supported models table so it doesn't depend on #44 landing first; whichever merges first, the other will need a small rebase.

---

Generated by the Sync Skills Cursor Automation. Please review before merging.

[broly-code-security-scanner]

Broly Security Scan

Note

✅ Clean scan
No vulnerabilities detected in this PR.

Note

Re-scan this PR anytime with /broly scan — useful after /broly undismiss, or to refresh findings without a new push.

Broly — SAST (zai-org/GLM-5.2) · Secrets · SCA · GH Actions (zizmor) · Containers · SBOM · Powered by Together AI