ellipticmarketing/modelrelay

Local router that benchmarks free coding models across providers and forwards requests to the best available model. Compatible with Opencode and Openclaw

What it solves

modelrelay is a local router that provides free access to a wide variety of AI coding models. It eliminates the need to pay for multiple API subscriptions by benchmarking and automatically routing requests to the best available free model across various providers.

How it works

It acts as an OpenAI-compatible proxy. When a request is sent to the auto-fastest endpoint or a specific model group, the router evaluates a Quality of Service (QoS) score based on a blend of the model's intelligence percentile and its recent average latency. It can route requests based on tags (e.g., coding, reasoning) or minimum context window requirements (min_ctx).

Who it’s for

Developers and AI coding tool users who want to access high-performance LLMs for free without managing multiple API keys or switching between providers manually.

Highlights

  • OpenAI-Compatible: Works as a drop-in replacement for existing tools and scripts.
  • Auto-Routing: Automatically selects the fastest and most capable model based on real-time benchmarks.
  • Broad Support: Access to 80+ models from 12+ providers including NVIDIA, Groq, and Google.
  • Flexible Routing: Supports routing by capability tags (e.g., tag:coding) and minimum context window filters.
  • Local Management: Includes a Web UI for managing providers, pinning models, and configuring QoS settings.

Related

  • Project
    Continuum-AI-Corp/OrcaRouter-Lite

    A self-hosted, OpenAI-compatible LLM router that automatically routes requests to the cheapest capable model across multiple providers.

  • Project
    romgX/openrelay

    A local AI quota aggregator and proxy that lets you use AI subscriptions from various providers (like Claude Pro or Cursor) across any AI tool or IDE.

  • Project
    tashfeenahmed/freellmapi

    A unified OpenAI-compatible API gateway that aggregates free tiers from dozens of LLM providers into a single endpoint with smart routing and failover.

  • Project
    workweave/router

    A drop-in LLM proxy that automatically routes every request to the best-performing model using a local embedder and cluster scoring.

  • Project
    decolua/9router

    A smart AI router and proxy that saves 20-40% of tokens and provides automatic fallback between subscription, cheap, and free AI models for coding tools.