Hyperfusion

What it does

Operation

POST /v1/chat/completions — inference, medium latency
OpenAI-compatible chat completions over DeepSeek V4 Flash, Qwen3.8-27B or MiniMax M2.7. The model is required. Price is a flat amount in one of two bands (prompt under or over ~4,000 estimated tokens), set by the model; there is no peak-hour surcharge. Additional OpenAI parameters (temperature, max_tokens, top_p, frequency_penalty, presence_penalty, stop, etc.) are forwarded. The optional metered (`upto`) accept authorizes the prompt plus your `max_tokens` as a ceiling and settles the measured input and output tokens. Streaming is not supported on the paid route.

Endpoint

https://hyperfusion.x402.press/v1/chat/completions

What a call costs

Measured from the service's own 402 response on 2026-10-30T23:28:43Z — unpaid, first round only, nothing signed or settled. The exact amount is what a fixed call costs and is the price this page publishes; the quoted amount is always returned in the handshake, so a client never has to compute it.

deepseek-ai/DeepSeek-V4-Flash-0731 — metered ceiling, bound to the request shape it was measured with.

Qwen/Qwen3.8-27B — metered ceiling, bound to the request shape it was measured with.

MiniMaxAI/MiniMax-M2.7 — metered ceiling, bound to the request shape it was measured with.

A ceiling is only meaningful next to the max_tokens it was measured with. On Base, a metered call with max_tokens: 1000 tops out at deepseek-ai/DeepSeek-V4-Flash-0731 $0.00955 / Qwen/Qwen3.8-27B $0.00955 / MiniMaxAI/MiniMax-M2.7 $0.02595; the same request with no max_tokens authorises the model's whole output allowance and is the right-hand column — a worst case, never a price. Ethereum is two orders dearer because that figure carries the chain's own settle cost, not the model's. Avalanche offers no metered accept, so a call there is always the exact price.

These are exactly the seven chains the handshake answered for hyperfusion; nothing above is a fallback or an estimate.

Start here

Two rounds, no account: send the request, read the 402, sign the amount it quotes, send it again. The first call is free to discover — a client learns the exact price before it pays anything.

curl -sS https://hyperfusion.x402.press/v1/chat/completions \
  -H 'content-type: application/json' \
  -d '{"model": "deepseek-ai/DeepSeek-V4-Flash-0731",
       "max_tokens": 1000,
       "messages": [{"role": "user", "content": "hi"}]}'
# 402 Payment-Required: base64url JSON, one accept per chain and per scheme
# (exact = the fixed price above; upto = the ceiling for that max_tokens)

The full endpoint list, parameters and manifest are on the service's own host: https://hyperfusion.x402.press.

Free endpoints

No payment, no key — the parts a client needs before it decides to pay.

• GET /v1/models — List the three served models
• GET /hyperfusion/feedback — Return the 50 most recent feedback entries
• POST /hyperfusion/feedback — Submit feedback (response_id, rating 1-5, optional comment)
• GET /.well-known/x402.json — machine manifest
• GET /llms.txt — the same offer, written for a model
• GET /health — liveness

For an agent

• /services/hyperfusion/llms.txt — this page as text: price, operation, free endpoints
• /services/hyperfusion.json — the same data, structured, derived from catalog.json
• /catalog.json — every service, every paid operation, every chain price
• /.well-known/x402.json (service) — the service's own manifest, the authority on its endpoints
• HyperFusion: three open-weight models, and where the GPUs sit — x402.review, independent long-form