docs: PZERO config2.yaml openai base_url example

Feature description

MetaGPT already talks to any OpenAI-compatible host via api_type: openai + base_url in ~/.metagpt/config2.yaml (https://docs.deepwisdom.ai/main/en/guide/get_started/configuration/llm_api_configuration.html). People looking for prepaid, privacy-first inference still have to guess the host, key format, and catalog id.

A named PZERO example on that page would let users paste and run. No first-party MetaGPT provider required.

Your Feature

Docs-only. After metagpt --init-config, edit ~/.metagpt/config2.yaml:

llm:
  api_type: "openai"
  model: "deepseek-v4-flash"
  base_url: "https://api.pzero.studio/v1"
  api_key: "pzero_YOUR_KEY"

base_url is the /v1 root, not /v1/chat/completions. api_type stays openai (the compatibility layer). Pass the catalog id as-is. Do not use the deprecated key.yaml / OPENAI_API_BASE names.

Get a key: https://pzero.studio/agents (sign-in free; min top-up 1 USDC on Base; no starter credits). Live text ids: GET https://api.pzero.studio/v1/models (public, no key). Default landing model: deepseek-v4-flash.

Smoke against the same host MetaGPT will call:

curl -sS -X POST "https://api.pzero.studio/v1/chat/completions" \
  -H "Authorization: Bearer $PZERO_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"deepseek-v4-flash","messages":[{"role":"user","content":"Hello from PZERO"}],"stream":false}'

I am not asking for a bundled api_type: pzero. The existing openai base_url seam is enough. Do not start implementation on my behalf until the board says Ready; the yaml is for users today.

Every AI Credit is at least 20% off $1 list. Usage is paid. Search for pzero returned none.