Last updated: August 27, 2026
Codex CLI can be connected to the Logicc Responses endpoint using its own model provider. To get started, we recommend gpt-5.6-terra as a balanced standard model.
Beta: Compatibility depends on whether the selected Logicc route fully supports the Responses endpoint and the tool calls required by Codex. Check the integration before production use.
LOGICC_API_KEY as an environment variable.~/.codex/config.toml as the Responses provider./status and quota via API endpoint./v1/responsesmacOS, Linux or WSL:
curl -fsSL https://chatgpt.com/codex/install.sh | sh
Alternatively with npm:
npm install -g @openai/codex
codex --version
macOS, Linux or WSL:
export LOGICC_API_KEY="YOUR_LOGICC_API_KEY"
Windows PowerShell:
$env:LOGICC_API_KEY = "YOUR_LOGICC_API_KEY"
Do not store the key in config.toml, Git or public scripts.
Create or open ~/.codex/config.toml:
model = "gpt-5.6-terra"
model_provider = "logicc"
[model_providers.logicc]
name = "Logicc API"
base_url = "https://api.logicc.cloud/v1"
env_key = "LOGICC_API_KEY"
wire_api = "responses"
requires_openai_auth = false
On Windows the path is %USERPROFILE%\.codex\config.toml.
The provider must be defined in the user-level configuration file. A project-level .codex/config.toml should not override model_provider or model_providers. For custom providers, Codex only supports wire_api = "responses".
curl "https://api.logicc.cloud/v1/models" \
-H "Authorization: Bearer $LOGICC_API_KEY"
Only use enabled GPT models whose route supports Responses and tool calling.
cd /path/to/project
codex
Windows PowerShell:
Set-Location "C:\Path\To\Project"
codex
Check active configuration:
/status
Override the model for a single run:
codex --model gpt-5.6-luna
| Model | Model ID | Recommendation | Cost profile |
|---|---|---|---|
| GPT-5.6 Terra | gpt-5.6-terra |
Balanced standard | Medium |
| GPT-5.6 Luna | gpt-5.6-luna |
Routine tasks and lower consumption | Lower |
| GPT-5.6 Sol | gpt-5.6-sol |
Challenging tasks | High |
Other direct OpenAI IDs like gpt-5.1, gpt-5.2, gpt-5.4, gpt-5.4-mini and gpt-5.5 should only be used if the specific route supports /v1/responses and agentic tool calls.
Azure GPT models such as gpt-5, gpt-5-mini, gpt-5-nano, o3-mini, gpt-4o-mini, gpt-4.1-nano, gpt-4.1-mini and gpt-4o are also considered conditionally compatible with Codex and must be tested for each route.
| Model type | Status |
|---|---|
| GPT with full Responses and tool support | Recommended |
| GPT without confirmed Responses support | Conditionally compatible |
| Claude and Gemini | Not recommended for this integration |
| Audio, transcription, embedding and image models | Not suitable |
Codex Desktop and Codex CLI read the user-wide Codex configuration. However, an environment variable set in the terminal is not automatically available to a separately started desktop process. Provide the key through your environment's secret manager, or launch the desktop app from an environment where LOGICC_API_KEY is available.
The initial setup and error analysis should be done via the CLI. Support for your own providers in the desktop app may depend on the app version.
Codex can execute many model queries and tool steps for one task. Check the remaining budget regularly:
curl "https://api.logicc.cloud/api/governance/virtual-keys/quota" \
-H "Authorization: Bearer $LOGICC_API_KEY"
Budget, usage, and remaining budget are tracked in euros.
Codex works locally in the selected project and executes approved terminal commands on the computer. Prompts and relevant file or code snippets are transferred for model processing via the Logicc API. Web and MCP functions can generate additional data flows.
| Error | Resolution |
|---|---|
Environment variable not found |
Set LOGICC_API_KEY in the same terminal and restart Codex |
| ChatGPT login appears | requires_openai_auth = false, check provider name and configuration location |
404 at the Responses endpoint |
Set base_url to https://api.logicc.cloud/v1 and check Responses support of the route |
403 or model not found |
Check /v1/models and model availability |
429 |
Check quota and rate limit |
| Tool calls fail | Switch to a confirmed GPT Responses route |
If you have any questions, write to [email protected] with the subject API. Include the Codex version, operating system, model ID, HTTP status, timestamp, and complete error message. Do not include the API key.
Further documentation: