windfall-inference
bởi papa-raw
Web & Frontend DevelopmentTóm tắt
Spatially-routed LLM inference at $0.004/req.
Mô tả
Kỹ năng này làm gì
Windfall is an LLM inference gateway that routes requests to nodes running on the cheapest, cleanest energy. Every request costs $0.004 flat and produces a verifiable onchain attestation on Base. It exposes an OpenAI-compatible API, so existing code works with a URL swap.
Drops into any OpenAI-compatible codebase with two env var changes while adding energy-aware routing and onchain proof of inference that direct provider APIs don't offer.
Khi nào nên dùng
- Swap in a greener inference backend without rewriting agent code
- Run a cost-sensitive AI pipeline across 200+ models at flat per-request pricing
- Verify inference provenance on Base for compliance or auditing
- Pay per request from a Base wallet without managing API keys
- Optimize LLM routing between carbon intensity and cost in real time
Quy trình ví dụ
User asks: run my chatbot using the cheapest green model available
- 1
Set OPENAI_BASE_URL to https://windfall.ecofrontiers.xyz/v1 and OPENAI_API_KEY to the Windfall key
- 2
Send chat completions request with model set to 'auto' and mode set to 'greenest'
- 3
Windfall engagement classifier selects DeepSeek V3 on a low-carbon node
- 4
Response returns with X-Windfall-Mode, X-Windfall-Model, and X-Windfall-Cost headers
- 5
EAS attestation is written to Base recording node location, carbon intensity, and response hash
Chat response delivered with verifiable green inference record on Base
Yêu cầu
- WINDFALL_API_KEY environment variable (free at windfall.ecofrontiers.xyz/api/keys)
- Base wallet address to generate an API key
- ERC-8004 agent identity or Basename for 100 free requests (optional, anonymous keys get 25)
Cài đặt
clawhub install papa-raw/windfall-inferenceChạy lệnh CLI, hoặc dán prompt vào Claude Code, Cursor, Codex hay coding agent bạn đang dùng.