Updated

Abliterated model in Codex

Codex uses an abliterated model when ~/.codex/config.toml names a provider that serves the Responses format. It reads the models and their effort levels from the provider.

  1. 1

    Install Codex

    npmnpm install -g @openai/codex
    Homebrewbrew install --cask codex
  2. 2

    Set your key

    Create a key on Silk Compute's API keys page and add credits from $5. Keep it in an environment variable. env_key tells Codex which variable to read.

    Shellexport SILK_API_KEY="sk-..."
  3. 3

    Add the provider

    Paste the file below into ~/.codex/config.toml. To keep OpenAI as the default, save it as ~/.codex/silk.config.toml and run codex --profile silk.

  4. 4

    Run it

    Switch the model and effort with /model.

    Runcodex
config.toml
~/.codex/config.toml
model = "deepseek-v4.1-flash-1017-uncensored"
model_provider = "silkcompute"
web_search = "disabled"

[model_providers.silkcompute]
name = "Silk Compute"
base_url = "https://api.silkcompute.com"
env_key = "SILK_API_KEY"
model_catalog_url = "https://api.silkcompute.com/codex/models"

Example output

/model
  kimi-k3.1-pro-uncensored
  deepseek-v4.1-flash-1017-uncensored
  deepseek-v4.1-flash-1017-fast-uncensored
  glm-5.3-uncensored
  glm-5.3-flash-uncensored