Verified 🤝 Inference

Trust but verify.

A simple FastAPI proxy for chat/completitons with ed25519 signing and Bittensor metagraph integration

Miners can opt-in and proxy their LLM requests through a verifier, which will forward their request to the defined provider and respond back with a signed response + proof. The response with proof is submitted to the querying Validator who can independently verify the proof and adjust rewards if applicable.

Trusted Inference

Known Limitations

• local miners are excluded from this implementation
• ed25519 keys are ephemeral

Configure Environment

.env (server)

B64_PRIVATE_KEY=ed25519 key
DATABASE_URL=postgres db

/tests/.env (testing)

OPENROUTER_KEY=sk-or-v1-xxxxx
OPENAI_KEY=sk-xxxxx
HOTKEY=your-miner-hotkey

Run

uv sync
uv run uvicorn app.main:app

Test

start the server first then:

uv run pytest

Endpoints

Server runs on http://127.0.0.1:8000

• GET / - root
• GET /health - health
• GET /log - verified log
• GET /providers - provider pings
• GET /public_key - public key
• GET /is_verified - miner verified lookup
• POST /v1/chat/completions - Proxy endpoint

License

MIT