What Is OmniRoute? Complete Guide to One API for 290+ AI Models (2026)

Change one base_url in Claude Code, Cursor, or your own Agent and you can switch between Claude, GPT, Gemini, DeepSeek, Kimi, and hundreds of other models on a single OpenAI-compatible API—that is not exclusive to OpenRouter. OmniRoute is among the fastest-growing open-source AI gateways on GitHub in 2026: MIT-licensed, self-hosted, with a catalog of 290+ providers and 500+ models (90+ with a free tier). This guide covers what it is, your first successful API call, Claude Code wiring, and a production security checklist.

What OmniRoute is and what it solves

In short: OmniRoute is an LLM reverse proxy on your own machine. It exposes unified /v1/chat/completions (plus Anthropic /v1/messages, Gemini /v1beta/models, and more) upstream, and connects to your provider API keys, OAuth subscriptions, or free pools downstream.

OmniRoute vs. OpenRouter vs. LiteLLM

For per-token price tables, read our OpenRouter pricing comparison. For Agent request budgets on OmniRoute, see Auto Combo budget routing.

Install & start (Docker / npm)

Docker (recommended)

docker run -d --name omniroute \
  -p 20128:20128 \
  -v omniroute-data:/data \
  diegosouzapw/omniroute

Open http://localhost:20128 → /dashboard.

npm (quick local trial)

npm install -g omniroute
omniroute

Dashboard in four steps

1. Create an API Key—all clients use Authorization: Bearer <key>.
2. Connect providers—OAuth, paste API keys, or enable free no-auth pools (trial only).
3. Point clients at http://<host>:20128/v1.
4. Monitor usage—management routes without a key must return 401.

List all models: GET /v1/models

export OMNI_KEY="your-omniroute-api-key"
curl -s http://localhost:20128/v1/models \
  -H "Authorization: Bearer $OMNI_KEY" | jq '.data | length'

First API call (curl / Python / Node)

Use provider/model IDs. Missing prefix may be auto-added; mismatches return 400.

curl

curl -s http://localhost:20128/v1/chat/completions \
  -H "Authorization: Bearer $OMNI_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"anthropic/claude-sonnet-4-6","messages":[{"role":"user","content":"Explain OmniRoute in one sentence"}],"max_tokens":256}'

Python

from openai import OpenAI
client = OpenAI(api_key="your-key", base_url="http://localhost:20128/v1")
print(client.chat.completions.create(
    model="openai/gpt-5.4",
    messages=[{"role": "user", "content": "Hello"}],
).choices[0].message.content)

Streaming

curl -N http://localhost:20128/v1/chat/completions \
  -H "Authorization: Bearer $OMNI_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"deepseek/deepseek-chat","messages":[{"role":"user","content":"Count to 5"}],"stream":true}'

Anthropic Messages

Native Anthropic clients can use POST /v1/messages. Full endpoint list: API Reference.

auto/* intent aliases & fallback

Aliases like auto/best-coding pick a healthy model under quota—no client change when you swap providers. Use budget routing in production to avoid unwanted free-tier fallback.

curl -s http://localhost:20128/v1/chat/completions \
  -H "Authorization: Bearer $OMNI_KEY" -H "Content-Type: application/json" \
  -d '{"model":"auto/best-coding","messages":[{"role":"user","content":"quicksort"}]}'

Claude Code, Cursor & Copilot

Production & security checklist

• Upgrade to latest release (v3.8.50+); remove default passwords;
• Never expose port 20128 on the public internet without TLS;
• Per-member API keys, token limits, IP allowlists;
• Treat free pools as fallback only; backup /data volume.

FAQ

400 model not found?

Check provider/model spelling and provider connection in the dashboard. Try auto/best-free to validate the chain.

Ollama locally?

Yes—POST /v1/api/chat integrates local models into the same routing table.

Use with OpenRouter?

Yes—OmniRoute can treat OpenRouter as a downstream provider for a hybrid setup.