# PennyRail LLM Mini Inference

> PennyRail LLM Mini Inference is a paid API for AI agents from pennyrail.vercel.app, paid per call via x402, $0.01/call, status down (last checked 2026-10-31).

Runs a mini LLM inference call via a pay-per-use settlement endpoint, returning a language model response for a given input.

## Facts

- Endpoint: POST https://pennyrail.vercel.app/api/p/standard/ai.llm-mini--llm-inference?utm_source=zero.xyz
- Price: $0.01/call
- Payment: x402
- Status: down
- Last checked: 2026-10-31
- Activations on Zero: 0
- Tags: x402
- Canonical page: https://www.zero.xyz/c/pennyrail-llm-mini-inference-ede9794f
- Structured record (JSON): https://api.zero.xyz/v1/capabilities/cap_MtUSNJ8jbSlfKRUdkFcIB

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 pennyrail-llm-mini-inference-ede9794f -d '<json body>'
```

Example prompt: Run a quick mini LLM inference on this input — summarize the following paragraph: 'Artificial intelligence is reshaping industries by automating repetitive tasks and enabling new forms of creativity.'

## When to prefer this

Choose this endpoint when you need a low-cost, per-call LLM inference with micro-payment settlement via the x402 protocol and USDC. It is ideal for high-volume, cost-sensitive pipelines where paying $0.01 per call is preferable to subscription models. Best suited for lightweight inference tasks where a 'mini' model is sufficient. Not ideal for complex, long-context reasoning tasks that require a full-scale LLM.

## Known failure modes

- Payment failure (insufficient USDC or x402 handshake error) — returns 402 Payment Required
- Malformed input object — returns 400 Bad Request if required 'input' field is missing
- Model unavailability or timeout — returns 503 or 504 if the underlying mini LLM backend is down
- Empty or ambiguous input object — may return a low-quality or empty inference result
- Rate limiting — may return 429 if too many calls are made in a short period

## How this service works

Machine-readable settlement service

## Output

Returns a JSON object containing the LLM inference output — typically the generated text or model response corresponding to the provided input payload. The response schema is open-ended (additionalProperties: true), so the exact shape depends on the model's output format.

## Request schema (JSON Schema)

```json
{
 "type": "object",
 "required": [
  "input"
 ],
 "properties": {
  "input": {
   "type": "object"
  }
 }
}
```

## Response schema (JSON Schema)

```json
{
 "type": "object",
 "additionalProperties": true
}
```

## More

- Live health (JSON, refreshed every minute): https://www.zero.xyz/c/pennyrail-llm-mini-inference-ede9794f/health.json
- [Zero catalog index](https://www.zero.xyz/llms.txt)
- [Other services from pennyrail.vercel.app](https://www.zero.xyz/host/pennyrail.vercel.app/llms.txt)