# OpenCovenant Compute Inference Chat Completions

> OpenCovenant Compute Inference Chat Completions is a paid API for AI agents from compute-inference.opencovenant.org, paid per call via x402, $0.014304/call, status unknown (last checked 2026-10-28).

Runs LLM text inference via a chat completions endpoint and returns a signed receipt from the operator who processed the request.

## Facts

- Endpoint: POST https://compute-inference.opencovenant.org/v1/chat/completions?utm_source=zero.xyz
- Price: $0.014304/call
- Payment: x402
- Status: unknown
- Last checked: 2026-10-28
- Activations on Zero: 0
- Tags: x402
- Canonical page: https://www.zero.xyz/c/opencovenant-compute-inference-chat-completions-ff15c1de
- Structured record (JSON): https://api.zero.xyz/v1/capabilities/cap_YnMS6WznK7JQzJ_R2lPP7

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 opencovenant-compute-inference-chat-completions-ff15c1de -d '<json body>'
```

Example prompt: Send this conversation to the OpenCovenant compute inference endpoint and get back a response with a signed operator receipt: system prompt is 'You are a helpful assistant', and the user message is 'Summarize the key differences between TCP and UDP in two sentences.'

## When to prefer this

Choose this endpoint when you need LLM text inference combined with cryptographic accountability — specifically when your use case requires a verifiable, operator-signed receipt proving who ran the inference. Ideal for compliance-sensitive workflows, auditable agentic pipelines, or any scenario where proof of compute provenance matters. Also suitable when you want pay-per-call LLM access via x402 micropayments without a subscription.

## Known failure modes

- Payment failure if x402 USDC balance is insufficient or wallet is not configured
- Invalid request format if messages array is malformed or missing required fields
- Model unavailability if the underlying compute node is offline
- Rate limiting or capacity errors during peak load
- Receipt signing failure if operator key is unavailable
- Timeout if inference takes too long for complex prompts

## How this service works

Text inference for AI agents, with a receipt signed by the operator who ran it.

## Output

Returns a chat completion response in OpenAI-compatible format containing the assistant's generated message, along with a cryptographic receipt signed by the operator who executed the inference, providing verifiable provenance of who processed the request.

## More

- Live health (JSON, refreshed every minute): https://www.zero.xyz/c/opencovenant-compute-inference-chat-completions-ff15c1de/health.json
- [Zero catalog index](https://www.zero.xyz/llms.txt)
- [Other services from compute-inference.opencovenant.org](https://www.zero.xyz/host/compute-inference.opencovenant.org/llms.txt)