Skip to content

tachikomared/nvidiarouter

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

2 Commits
 
 
 
 
 
 

Repository files navigation

NVIDIA Router for OpenClaw

High-Performance Intelligent Routing for NVIDIA-Hosted Models

NVIDIA Router is an open-source, intelligent routing layer for OpenClaw. It acts as the "brain" for your AI agents, automatically selecting the optimal model for every request based on task complexity, cost, and specific capabilities.

What makes it different?

  • Intelligent Tiering (SIMPLE to VISION): No more "one model fits all." Automatically route tasks to the best-fit model:
    • SIMPLE: Low-cost, high-speed tasks (e.g., greetings, simple translations).
    • MEDIUM: Mid-range tasks (e.g., creative writing, standard analysis).
    • COMPLEX: Deep reasoning and architectural planning.
    • REASONING: Advanced chain-of-thought (DeepSeek, GLM, Moonshot).
    • VISION: Multimodal image processing and scene understanding.
  • Leverage NVIDIA Free-Tier: Maximize the generous free-tier limits provided by the NVIDIA API catalog without proprietary middleman fees.
  • Intelligent Fallback Chains: If a primary high-performance model hits capacity, the router seamlessly switches to high-performance alternatives without failing your agentic workflow.
  • BYOK (Bring Your Own Key): Full ownership. Your API keys, your control.

Setup

  1. Get your API Key: build.nvidia.com
  2. Configuration: Add the plugin to your openclaw.json and start routing.
  3. Run: Launch your OpenClaw gateway and route prompts to nvidiarouter/auto.

Why this changes everything

Tired of paying premium fees for simple tasks or dealing with locked AI workflows? NVIDIA Router gives you agentic autonomy:

  • Cost-Optimization: Dynamically prioritize efficiency to get more agent runs per free-tier credit.
  • Workflow Resilience: Intelligent automatic routing ensures you only pay for complexity when it's truly needed.
  • Open Source Autonomy: Host your own smart routing layer locally. No proprietary SaaS dependencies.

About

Intelligent, high-performance local model router for OpenClaw. Routes prompts across NVIDIA-hosted models (Mistral, Nemotron, DeepSeek, etc.) using capability-based tiering to maximize free-tier efficiency.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors