AI inference,
routed openly
Decentralized inference gateway powered by community GPUs. Transparent billing, Lightning payments, 100% open source.
Decentralized inference gateway powered by community GPUs. Transparent billing, Lightning payments, 100% open source.
Everything a closed gateway does, but you can read every line of code.
Gateway, billing, routing — all public on GitHub. Audit the code that handles your money.
Providers earn sats automatically. Users pay per token. No accounts, no credit cards, no middlemen.
Requests are routed to the best available node by model, VRAM, queue depth, and region.
Per-token pricing, 5% flat fee, every job logged in SQLite. No hidden charges.
Drop-in replacement for the OpenAI API. Change one line — your base URL — and it works.
No accounts, no emails, no tracking. We don't log your prompts or responses. Provider nodes never know who you are. Your API key is your only identity.
Anyone with a GPU can join as a provider. Earn Bitcoin from idle compute. No lock-in.
RouteCat acts as a privacy proxy between you and the inference nodes. We don't store, log, or inspect your prompts or responses. Provider nodes never see your identity — they only receive the raw inference request. No emails, no sign-ups, no cookies. Your API key is the only link to your account, and it's generated locally with no personal data attached.
Pay per token. Providers keep 95%, gateway takes 5%.
| Model | Input (per 1M tokens) | Output (per 1M tokens) |
|---|---|---|
| Loading pricing... | ||
OpenAI-compatible. Works with any SDK or curl.
Get an API key. Top up with Lightning. Check your balance.
Create a free API key to try the playground (10 requests/day), or enter an existing key.
Already have a key?
⚠ Save your API Key now!
This key will not be shown again. If you lose it, you lose access to any funds deposited. Copy it and store it safely.
Try it live. No sign-up needed for a quick test.
Got a GPU? Earn Bitcoin by serving inference.
RouteCat acts as a proxy between you and the inference nodes. Your prompts and responses are forwarded in real time and never stored. Provider nodes only see the raw model request — they don't know who you are. No emails, no cookies, no tracking. Your API key is your only identity, and it carries no personal information.
Three ways: (1) The code is open source — read it yourself. (2) The /v1/stats endpoint shows the exact git commit hash of the running binary — you can rebuild from source and compare. (3) The /v1/audit endpoint shows all completed jobs with token counts, earnings, and fees — no hidden charges.
Providers set a Lightning address and a payout threshold (in sats). When their earned balance exceeds the threshold, the gateway automatically sends a Lightning payment. Providers keep 95% of every job — the gateway takes a flat 5% fee. All payouts are verifiable on-chain via the payment hash.
Create an API key (free, 10 playground requests/day). For API access, top up your balance by paying a Lightning invoice — any amount from 10 to 100,000 sats. Your balance is deducted per-token based on the model's pricing. No subscriptions, no minimum commitments.
Your API key is shown only once when you create it. If you lose it, you lose access to any remaining balance. There is no recovery mechanism — this is by design, as we don't collect any personal data to verify your identity. Always save your key securely.
Yes. Set base_url="https://route.cat/v1" and your API key in any OpenAI-compatible client (Python, Node.js, curl). Streaming, chat completions, and model listing all work the same way.
Questions, ideas, or just want to say hi? Join the conversation.
Announcements, updates, and new features. Follow to stay in the loop.
Ask questions, report issues, share ideas, or help others. Open to everyone.
Found a bug or have an improvement? Open an issue or submit a pull request on GitHub. Every contribution helps.