# Anthropic Claude Chat Completions (OpenAI-Compatible)

> Anthropic Claude Chat Completions (OpenAI-Compatible) is a paid API for AI agents from Anthropic at anthropic.mpp.tempo.xyz, paid per call via MPP, $0.01025/call, not verified (last probed 2026-10-11); live alternatives: agent402.tools LLM Inference Proxy ($0.01) and Bankr LLM Gateway (free).

Agents call it through Zero, which pays per call: no API key and no wallet of your own. New accounts start with $5.

Live alternatives for the same use case:
- [agent402.tools LLM Inference Proxy](https://www.zero.xyz/c/agent402-tools-llm-inference-proxy-gpt-4o-mini-0d820d98) ($0.01)
- [Bankr LLM Gateway](https://www.zero.xyz/c/bankr-4f042050) (free)

Sends an OpenAI-format chat completion request that is auto-converted and fulfilled by Anthropic Claude models (Sonnet, Opus, Haiku)

## Facts

- Endpoint: POST https://anthropic.mpp.tempo.xyz/v1/chat/completions?utm_source=zero.xyz
- Price: $0.01025/call
- Payment: MPP
- Status: unknown
- Last checked: 2026-10-11
- Last successful call: 2026-05-29
- Success rate: 78% of calls made through Zero
- Rating: 2.0 / 5 from 4 reviews
- Activations on Zero: 45
- Provider: Anthropic
- Docs: https://docs.anthropic.com
- Website: https://anthropic.com
- Tags: llm, claude, sonnet, opus, haiku, chat, ai
- Canonical page: https://www.zero.xyz/c/anthropic-completions-3be6ac9d
- Structured record (JSON): https://api.zero.xyz/v1/capabilities/cap_7MHyL-bhbn-bqG24hmXJs

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 anthropic-completions-3be6ac9d -d '<json body>'
```

Example prompt: Using the OpenAI-compatible Claude endpoint, send this conversation to claude-3-5-sonnet-20241022: system prompt 'You are a helpful assistant', user message 'Explain the difference between supervised and unsupervised learning in 3 sentences', with max_tokens set to 300.

## When to prefer this

Choose this endpoint when you already have OpenAI SDK client code and want to route requests to Claude models without rewriting your integration. Ideal for agents that need drop-in Claude access using the familiar /v1/chat/completions path, avoiding the need to learn Anthropic's native Messages API format. Also prefer this when your orchestration framework is hardcoded to OpenAI-style schemas.

## Known failure modes

- Invalid or unsupported model name returns 400 error
- Malformed messages array (missing role or content) causes validation error
- Exceeding context window length for selected Claude model returns 400
- Insufficient USDC balance or payment failure returns 402
- Rate limiting or overload returns 429 or 503
- Streaming requested but not supported may return error or fall back

## How this service works

OpenAI-compatible chat completions (auto-converted to Anthropic format)

## Output

Returns an OpenAI-format chat completion response object containing the assistant's generated message content, finish reason (e.g. 'stop'), token usage counts (prompt, completion, total), and model identifier — all in the standard OpenAI ChatCompletion JSON shape despite being fulfilled by Claude.

## More

- Live health (JSON, refreshed every minute): https://www.zero.xyz/c/anthropic-completions-3be6ac9d/health.json
- [Zero catalog index](https://www.zero.xyz/llms.txt)
- [Other services from anthropic.mpp.tempo.xyz](https://www.zero.xyz/host/anthropic.mpp.tempo.xyz/llms.txt)