Codex CLI on uncensored Kimi K3.1 Pro

Codex CLI speaks only the OpenAI Responses API. Wild West serves that format at /v1/responses, so Codex can drive Kimi K3.1 Pro Outlaw with a provider entry and no shim.

The provider block

Add this to ~/.codex/config.toml:

model = "kimi-k3.1-pro-outlaw"
model_provider = "wildwest"

[model_providers.wildwest]
name = "Wild West"
base_url = "https://wildwestapi.com/v1"
env_key = "WILDWEST_API_KEY"
wire_api = "responses"

Then export the key and start Codex:

export WILDWEST_API_KEY=sk-ww-...
codex

The base URL keeps /v1 because the Responses client appends only /responses. The same block with codex exec runs it headless.

Why this model suits Codex

Codex edits files and runs commands only through function calls, so a model that cannot make one cannot do anything. Kimi K3.1 Pro Outlaw calls tools, reasons before it acts, reads 1,048,576 tokens of a repository and can write up to 943,718 in one reply. It is the strongest model on the line, so it suits the sessions where the plan matters most; Slingshooter is the cheaper choice for routine ones.

What is translated, and what is not

A Responses request becomes a chat-completions body on the way out and the answer becomes a Response object on the way back, with the real event sequence when streaming. Function tools you define work normally.

• previous_response_id is refused with a clear error. Codex sends the full input every turn, so it does not need it.
• OpenAI's hosted tools (web_search, file_search, computer_use) run inside OpenAI's own stack, so they are not available.

The general setup, for any model on the line, is in the Codex CLI quickstart.

FAQ

Related