Integrations

Codex

Codex CLI speaks the Responses API. Add the gateway as a model provider and point Codex at it.

Configuration

Put this in ~/.codex/config.toml, or %USERPROFILE%\.codex\config.toml on Windows.

~/.codex/config.toml

model = "gpt-5.6-sol"
model_provider = "overagent"

[model_providers.overagent]
name = "Overagent"
base_url = "https://api.overagent.cc/v1"
env_key = "OVERAGENT_API_KEY"
wire_api = "responses"

export OVERAGENT_API_KEY=sk-oa_...
codex

The three settings that matter

base_url includes /v1 — the opposite of the Anthropic side.
wire_api must be responses, so Codex calls POST /v1/responses rather than the chat endpoint.
env_key names the environment variable holding your key. Codex reads it at start-up.

Choosing a model

model takes any id from the catalog, not only GPT-family ones. Codex does not care which vendor is behind the id it is given.