windfall-inference（意外收获推断）

Windfall Inference

Spatially-routed LLM inference gateway for AI agents on Base. Routes every request to the cheapest model on the cleanest energy.

Setup

Set WINDFALL_API_KEY in your environment. Get one free at:

curl -X POST https://windfall.ecofrontiers.xyz/api/keys \
  -H "Content-Type: application/json" \
  -d '{"wallet_address": "YOUR_WALLET"}'

Keys with an ERC-8004 agent identity or Basename get 100 free requests. Anonymous keys get 25.

Usage

Use as an OpenAI-compatible endpoint. Set these two environment variables:

OPENAI_BASE_URL=https://windfall.ecofrontiers.xyz/v1
OPENAI_API_KEY=wf_YOUR_KEY

Or call directly:

curl https://windfall.ecofrontiers.xyz/v1/chat/completions \
  -H "Authorization: Bearer $WINDFALL_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "auto",
    "messages": [{"role": "user", "content": "Hello"}]
  }'

x402 Payment (No API Key Needed)

Windfall supports the x402 payment protocol. Any agent with a Base wallet can pay per-request without creating an API key. Send a request without auth and the server returns HTTP 402 with a PAYMENT-REQUIRED header containing USDC payment instructions on Base. Your x402 client handles the rest.

Routing Modes

• greenest (default) — lowest carbon intensity
• cheapest — lowest energy cost
• balanced — Pareto-weighted cost + carbon

Set via "mode" in the request body or X-Routing-Mode header.

Models

Default: DeepSeek V3 (auto-selected by engagement classifier). Override with "model" field. 200+ models available via OpenRouter.

Pricing

• Standard: $0.004/request
• Premium (Claude, GPT-4): $0.008/request
• Green surcharge: +10% for greenest mode
• Cache hits: free

Response Headers

Every response includes:

• X-Windfall-Cache — HIT or MISS
• X-Windfall-Mode — routing mode used
• X-Windfall-Model — model that handled the request
• X-Windfall-Node — node that executed the request
• X-Windfall-Cost — cost charged
• X-Windfall-Saved — savings vs direct API (cache hits)

Onchain Attestations

Every inference call produces a verifiable EAS attestation on Base with: node location, energy price, carbon intensity, model used, and response hash. Query at base.easscan.org.