Reference/llm.peck.to
llm.peck.to
Multi-provider LLM gateway with per-request BSV micropayments.
What it is
OpenAI-compatible `/v1/chat/completions` router over Gemini, OpenAI, Anthropic and Ollama. Picks the cheapest provider per request, bills the caller via HTTP 402 + BRC-42, and applies a 1.05× margin by default. Lets agents pay for inference in BSV without holding an API key for every provider.
When to use it
An agent needs LLM inference and should pay per-token in BSV instead of carrying a provider API key.
Clients that wrap it
Most apps talk to this service through one of these packages rather than hitting the HTTP surface directly.
openai (any OpenAI-compatible client)
Deployment
Repo: llm-gateway · Framework: go · Protocol: OpenAI-compatible / HTTP 402
Endpoints
9 routes scanned from source.
| Method | Path | Source |
|---|---|---|
| GET | X-Routing-Strategy | internal/gateway/handler.go:134 |
| GET | X-Forwarded-For | internal/middleware/ratelimit.go:58 |
| GET | X-Channel-ID | internal/middleware/auth.go:51 |
| GET | X-Channel-ID | internal/middleware/auth.go:64 |
| GET | Authorization | internal/middleware/auth.go:81 |
| ANY | /v1/chat/completions | cmd/gateway/main.go:69 |
| ANY | /v1/models | cmd/gateway/main.go:70 |
| ANY | /v1/estimate | cmd/gateway/main.go:71 |
| ANY | /health | cmd/gateway/main.go:72 |