Chat with any inference.net model straight from your terminal. The SSH username is the model id and your API key is the password.
$ ssh kimi-k3@ssh.inference.cool password: <your sk-inference-… API key>
Or skip the prompt by embedding the key in the username:
$ ssh "kimi-k3:$INFERENCE_API_KEY@ssh.inference.cool"
── inference.net ssh bridge ── model kimi-k3 · balance $498.20 type a message · /help for commands you › why is the sky blue? kimi-k3 › Rayleigh scattering: shorter wavelengths… 2.4s · 132 in / 458 out tok · $0.0072
Responses stream token-by-token; each answer ends with a stats line showing elapsed time, token counts, and estimated cost. Your team's credit balance is shown at login.
/help | show available commands |
/balance | show remaining credit balance |
/clear | reset the conversation |
/quit | disconnect (also exit or Ctrl+D) |
Ctrl+C | stop a response mid-stream |
Any model id from api.inference.net/v1/models works as the username — ssh in with a wrong one and the bridge lists what's available. Create API keys in the inference.net dashboard.