Codex CLI with an uncensored model

Where the settings are

Codex CLI reads ~/.codex/config.toml. A custom provider is a [model_providers.<id>] table there: a display name, a base URL, the environment variable that holds the key and the wire API. Codex speaks the Responses API only, and every build on the models page answers /v1/responses, so the one table covers them all. A profile is a separate file next to it, selected with --profile.

Steps

1. Add the table to config.toml. The ids openai, ollama and lmstudio are reserved, so the table needs its own name:

2. Create a key in the console, export it in the shell that runs Codex, then start it:

3. Send a prompt. The request shows on the console’s usage page within seconds, with its tokens and its charge. For one run on another id: codex --model glm-5.3-flash-uncensored.

What each setting does

model_provider

uncensoredllm

selects the table by its id; openai is the default

name

UncensoredLLM

the display name Codex shows for the provider

base_url

https://api.uncensoredllm.ai/v1

with the /v1; Codex adds /responses

env_key

UNCENSOREDLLM_API_KEY

the variable that holds the whole sk-ullm- key; it never goes in the file

wire_api

responses

the only value Codex accepts, and the default, so the line is optional

model

glm-5.3-uncensored

any id from the models page, exactly as it is spelled there

A profile per model

A profile is a file named ~/.codex/<name>.config.toml holding only the keys that differ from config.toml. Start Codex with --profile <name> and it layers the file on top. Guides written for older Codex versions show [profiles.<name>] tables inside config.toml; the reference now documents the separate files.

With a custom provider active, the /model menu lists Codex’s built-in models, not your provider’s, and picking one rewrites the active profile’s model line (openai/codex issue #35487, read 2026-11-06). An id that is not on the models page then returns 404. To undo it, set model back in the file; to switch without the menu, use --profile or --model.

Picking a model

Every agent turn resends the session and the tool output, so the input price carries most of the bill. The cost column is one turn of 30,000 tokens in and 1,000 out at the catalog prices on 2026-11-09; a session that repeats the same prefix bills the cached input tokens at the model’s cached rate where it has one.

For coding specifically, the uncensored coding AI guide compares the code builds and costs a full coding day on each.

Check it

Check the key and the Responses path outside Codex first. -i prints the headers: x-uplink-request-id names the request, and on a non-streamed answer x-uplink-charge is what it cost in USD.

Errors and fixes

• 401 invalid_api_key: the key is wrong or revoked, or UNCENSOREDLLM_API_KEY is empty in the shell that started Codex. Export it and start Codex again.
• 402 insufficient_balance: the balance does not cover the request. Top up in the console. 402 monthly_limit_reached means the key hit the monthly spend limit you set on it.
• 404 model_not_found: the model id is not on the models page, often after the /model menu rewrote a profile. Fix the model line.
• 429 rate_limited: the key hit its rate limit. The response carries retry-after.
• wire_api = "chat" from an older guide: Codex no longer accepts it. Set responses or delete the line.
• A profile has no effect: it is still a [profiles.<name>] table in config.toml. Move its keys into ~/.codex/<name>.config.toml.
• The table is not used: its id is a reserved name, or model_provider names a different id.

Sources

Keys and values follow Codex’s Configuration Reference, read 2026-11-06. Prices are from the live catalog at render.

Questions