NeuronPool

Docs

For developers

Use open models (Llama, Qwen, Gemma, gpt-oss, …) through the OpenAI or Anthropic API you already know. Authorize with an API key; a pool key keeps traffic inside its pool.

base_url https://neuronpool.damnknee.workers.dev/v1

client.py
  1. # Open-weight models on the pool (Llama, Qwen, Gemma, …).
  2. # OpenAI SDK is one client; Anthropic /v1/messages uses the same key.
  3. from openai import OpenAI
  4. client = OpenAI(
  5. base_url="https://neuronpool.damnknee.workers.dev/v1",
  6. api_key="sk-neuronpool-…",
  7. )
  8. client.chat.completions.create(
  9. model="neuronpool-tiny-chat",
  10. messages=[{"role":"user","content":"hello"}],
  11. )

Routes

POST /v1/chat/completionsOpenAI chat (streaming supported)
POST /v1/completionsOpenAI legacy completions
POST /v1/responsesOpenAI Responses (Cline, Cursor Agent)
POST /v1/messagesAnthropic Messages
POST /v1/embeddingsEmbeddings
POST /v1/images/generationsImage generation
POST /v1/audio/transcriptionsSpeech to text
GET /.well-known/neuronpool.jsonDiscovery document

Clients

Coding agents like Claude Code, Cline, and Continue work against a pool — see the setup guide. Cursor Chat needs a signed-in Cursor account. Or just try a model in the playground.

Hosts

Hosts run the CLI agent: curl -fsSL https://neuronpool.damnknee.workers.dev/install.sh | sh, then neuronpool login. Public jobs need a released agent build and a passing key challenge every 5 minutes — see trust levels.

Billing

Credits charge only on a successful job. Host cash-out is $10 minimum after a 7-day holdback. Settled Stripe transfers are sent; bank arrival is not tracked. Stripe is test mode until live keys are set.

Security

Trust levels. Verify a blessed build. Verifiable inference. On-chain receipt proofs stay local until Base RPC is configured.

Guides

Trust levels · Verifiable inference · Coding agents on a pool · Pool fair-use limits · Self-hosting the coordinator · Benchmark methodology