The Free Gateway to Every AI Model
One OpenAI-compatible endpoint. Intelligent routing across tested free models, automatic failover, and zero subscriptions. Point your SDK and build.
Engineered for Free AI
Core infrastructure built to make free-tier models behave like production ones.
Unified Model Access
Chat, vision, and reasoning through one OpenAI-compatible endpoint. Switch models with a single string — no SDK changes.
Model catalog →Automatic Failover
When a free model hits its rate limit, requests reroute to the next healthy model in the combo within milliseconds.
How routing works →Quota Preflight
Upstream quota is verified before every dispatch, so requests never leave the gateway just to die on a 429.
Preflight specs →Stream Recovery
Interrupted streams reconnect transparently mid-generation, so long agent runs survive provider hiccups.
Resilience →Featured Models
9 tested free models on the free-openrouter combo
How It Works
Get up and running with your free key in under 2 minutes.
Create your API key
Sign in to the dashboard and generate a production key in seconds. No credit card, no plan picker.
Point your SDK
Swap the base URL to api.pesatrouter.com/v1. OpenAI-compatible — streaming and tool calling work natively.
Route your first request
Call the combo ID free-openrouter and let the gateway pick the healthiest free model for every turn.
Works with your Favorite Tools
Drop-in compatibility with leading AI coding assistants and CLI agents.
Zero-Downtime Free Routing
Free tiers are volatile — PesatRouter is built for it. Quota preflight, health-scored circuit breakers, and transparent stream recovery keep your agents running through upstream turbulence. Transient 503 chat_admission_busy responses are automatically retried and failover-safe.
Upstream quota is verified before a request leaves the gateway.
Requests stream through the healthiest free model in the combo.
Failing providers are tripped open instantly and traffic reroutes.
Interrupted streams reconnect transparently, mid-generation.
Frequently Asked Questions
Everything you need to know about PesatRouter models, routing, and integrations.
Yes. The free-openrouter combo runs entirely on free-tier models served through the gateway. There are no subscriptions, no top-ups, no per-token billing, and no expired balances.
Nine tested free models, failover-ordered: google/gemma-4-26b-a4b-it, poolside/laguna-s-2.1, nvidia/nemotron-3-nano-omni-30b-a3b-reasoning, nvidia/nemotron-nano-12b-v2-vl, openrouter/free, nvidia/nemotron-3-ultra-550b-a55b, nvidia/nemotron-3-super-120b-a12b, nvidia/nemotron-3.5-lightning, and nvidia/nemotron-3-nano-30b-a3b — with context windows from 128K up to 1M tokens.
Yes. Point any OpenAI SDK at https://api.pesatrouter.com/v1/chat/completions with your PESAT_API_KEY bearer token. Streaming and tool calling work natively with zero code changes.
The circuit breaker trips and requests reroute to the next healthy model in the combo within milliseconds. A 503 chat_admission_busy response means the upstream free tier is momentarily overloaded — it is transient and retried automatically.
Claude Code, OpenCode, Cursor, Aider, Codex, Cline, Roo Code, and Windsurf are first-class — plus basically any OpenAI-compatible tool, agent, or SDK. Just configure the base URL and your Pesat API key.
The dashboard lives at api.pesatrouter.com (login required) for key management. Live gateway status is public at api.pesatrouter.com/healthz.
PesatRouter runs on OmniRoute, an open-source multi-provider AI gateway with combo-based failover, circuit breakers, and provider health tracking. Explore the source at github.com/diegopw/omniroute.
Start Building in 2 Minutes
Create your key, point your SDK at the gateway, and route your first free-model request.