Model Gateway

Documentation Index

Fetch the complete documentation index at: https://docs.insforge.dev/llms.txt

Use this file to discover all available pages before exploring further.

Use the Model Gateway to call chat, streaming, and embedding models through one OpenAI-compatible endpoint. InsForge holds the provider keys, tracks usage per project, and routes traffic through OpenRouter, so your application code never sees Anthropic, OpenAI, or Mistral credentials directly.

​

Features

​

OpenAI-compatible API

Point any OpenAI SDK or openai-compatible library at https://<project>.insforge.dev/v1 and it works. /v1/chat/completions, /v1/embeddings, and /v1/models all behave like the upstream spec.

​

Streaming

Server-sent events for chat completions. Use the streaming endpoint the same way you would with OpenAI; the gateway forwards tokens as they arrive from the provider.

​

Embeddings

Generate dense vectors from any embedding model OpenRouter supports. Store the result in Postgres with pgvector for semantic search.

​

Per-project quotas

Each project carries its own rate limit and spend cap. Hit it, and the gateway returns a clean 429 instead of leaking provider quota state into your app.

​

Usage tracking

Every request is logged with model, token count, and cost. Query usage from the dashboard, CLI, or MCP — billing reconciles to OpenRouter’s invoice automatically.

​

Multi-provider routing

Switch between Anthropic, OpenAI, Mistral, Llama, Gemini, and dozens more by changing the model name in the request. Application code does not change.

​

Build with it

​

Next steps

• Set up the CLI to link your project (the recommended path).
• Browse the TypeScript SDK reference for chat and embedding patterns.