Tool Setup
Configure Codex App
Codex App reads the same local .codex configuration used by Codex CLI.
Last updated 2026-05-12 | Maintained by TokenPool documentation team
Codex App can use TokenPool through the same local .codex configuration used by Codex CLI. Configure ~/.codex/config.toml and ~/.codex/auth.json once, then Codex App can use your TokenPool API Key and OpenAI-compatible base_url.
Directory
macOS / Linux:
~/.codex
Windows:
%USERPROFILE%\.codex
config.toml
model_provider = "OpenAI"
model = "gpt-5.5"
review_model = "gpt-5.5"
model_reasoning_effort = "xhigh"
disable_response_storage = true
network_access = "enabled"
windows_wsl_setup_acknowledged = true
model_context_window = 1000000
model_auto_compact_token_limit = 900000
[model_providers.OpenAI]
name = "OpenAI"
base_url = "https://api.clawto.link"
wire_api = "responses"
requires_openai_auth = true
auth.json
{
"OPENAI_API_KEY": "<your TokenPool API Key>"
}
If GPT-5.5 is missing
macOS / Linux:
TOKENPOOL_CODEX_MODEL='gpt-5.5' TOKENPOOL_CODEX_SET_DEFAULT_MODEL='1' bash -c "$(curl -fsSL 'https://cdn.tokenpool.dev/scripts/codex-fix-gpt55-catalog.sh')"
Windows PowerShell:
$env:TOKENPOOL_CODEX_MODEL='gpt-5.5'; $env:TOKENPOOL_CODEX_SET_DEFAULT_MODEL='1'; irm 'https://cdn.tokenpool.dev/scripts/codex-fix-gpt55-catalog.ps1' | iex
Screenshot placeholder: Codex App model selector.
Verify Codex App
- Restart Codex App after editing local configuration files.
- Confirm the selected model is available.
- Run a small request before starting a large task.
- Check TokenPool usage and quota in the console.
FAQ
Does Codex App need a separate TokenPool key?
No. Codex App can use the same TokenPool API Key stored in ~/.codex/auth.json.
Does Codex App use the same base_url as Codex CLI?
Yes. Both tools read the same base_url from the local .codex configuration when configured this way.
What should I do if Codex App cannot see GPT-5.5?
Run the catalog fix command on this page, then restart Codex App and check the model selector again.