# fold · x402 LLM Inference

> fold · x402 LLM Inference is a paid API for AI agents from x402.fold.computer, paid per call via x402, $2/call, status down (last checked 2026-09-15).

Pay-per-call LLM completions via OpenRouter models, billed in USDC over x402 protocol

## Facts

- Endpoint: POST https://x402.fold.computer/llm
- Price: $2/call
- Payment: x402
- Status: down
- Last checked: 2026-09-15
- Activations on Zero: 0
- Tags: x402
- Canonical page: https://www.zero.xyz/c/fold-x402-llm-inference-bbf4246b
- Structured record (JSON): https://api.zero.xyz/v1/capabilities/cap_JogpWGChep1EjhBFeTCqJ

Status and success rate cover calls made through Zero and Zero's own probes. Third-party monitors may report differently.

## How to call it through Zero

Zero handles the 402 payment challenge and records the run. With the Zero CLI installed (`npm i -g @zeroxyz/cli`):

```sh
zero fetch --capability fold-x402-llm-inference-bbf4246b -d '<json body>'
```

Example prompt: Use the fold x402 LLM endpoint with the anthropic/claude-opus-4-5 model to answer this question: 'What are the top 3 DeFi protocols by TVL right now?' — keep it under 300 tokens.

## When to prefer this

Choose this endpoint when your agent needs to call an LLM and pay per-request in USDC via the x402 protocol — especially useful for onchain or crypto-native agents that don't want recurring subscriptions. Prefer this over direct OpenAI/Anthropic APIs when you need crypto-native micropayment billing or want access to 300+ OpenRouter models through a single x402-compatible endpoint.

## Known failure modes

- Payment insufficient or missing — 402 response requiring x402 payment header
- Invalid model ID — returns error if model string doesn't match an available OpenRouter model
- Malformed messages array — 400 if messages don't follow OpenAI chat format
- Model rate limit or unavailability — upstream OpenRouter error propagated
- Insufficient USDC balance — payment processor rejects transaction

## How this service works

Machine-readable market intelligence for onchain agents, served over x402.

## Output

An OpenAI-format chat completion object (choices, usage, finish_reason) alongside x402 cost metadata showing how much USDC was charged for the request.

## Request schema (JSON Schema)

```json
{
 "type": "object",
 "required": [
  "model",
  "messages"
 ],
 "properties": {
  "model": {
   "type": "string",
   "description": "OpenRouter model id, e.g. openai/gpt-5.5, anthropic/claude-opus-4.8. See GET /llm/models for all 339."
  },
  "messages": {
   "type": "array",
   "description": "OpenAI-format chat messages"
  },
  "max_tokens": {
   "type": "number"
  },
  "temperature": {
   "type": "number"
  }
 }
}
```

## Response schema (JSON Schema)

```json
{
 "type": "object",
 "required": [
  "status",
  "data"
 ],
 "properties": {
  "data": {
   "type": "object",
   "description": "OpenAI-format completion + x402 cost info"
  },
  "status": {
   "type": "string"
  }
 }
}
```

## More

- Live health (JSON, refreshed every minute): https://www.zero.xyz/c/fold-x402-llm-inference-bbf4246b/health.json
- [Zero catalog index](https://www.zero.xyz/llms.txt)
- [Other services from x402.fold.computer](https://www.zero.xyz/host/x402.fold.computer/llms.txt)