Aimesh
Loading Aimesh
Preparing the next screen, account state, and workspace data.
Preparing secure workspace
Refreshing the latest data
Rendering the next screen
Preparing the next screen, account state, and workspace data.
Follow the 4 steps below: create your Aimesh API key, install Codex, paste the config files, then restart Codex and verify the connection.
Use this base URL in your Codex provider config.
Paste your user API key into auth.json as OPENAI_API_KEY.
Each user has one active API key at a time. Regenerating revokes the previous key.
Start in the dashboard, because you need a fresh active API key before you configure any Codex client.
Choose the Codex client you actually use. The same Aimesh config works across CLI, the VS Code extension, and the Codex App.
Open your local .codex directory and place both required files there: config.toml and auth.json.
After saving the files, restart your client and run a simple test so you know the proxy config is actually being used.
Start in the dashboard, because you need a fresh active API key before you configure any Codex client.
Use this base URL in your Codex provider config.
Paste your user API key into auth.json as OPENAI_API_KEY.
Each user has one active API key at a time. Regenerating revokes the previous key.
Choose the Codex client you actually use. The same Aimesh config works across CLI, the VS Code extension, and the Codex App.
Install the latest CLI globally if you want terminal-first usage.
npm i -g @openai/codex@latestInstall the official OpenAI extension and let it read the same config directory.
Open your local .codex directory and place both required files there: config.toml and auth.json.
start "" "%USERPROFILE%\\.codex"These templates follow the reference flow but swap in your Aimesh proxy endpoint instead of the external service URL.
model_provider = "aimesh"
model = "gpt-5.4"
model_reasoning_effort = "high"
network_access = "enabled"
disable_response_storage = true
windows_wsl_setup_acknowledged = true
model_verbosity = "medium"
[model_providers.aimesh]
name = "aimesh"
base_url = "https://aimesh.vercel.app/api/v1"
wire_api = "responses"
requires_openai_auth = trueAfter saving the files, restart your client and run a simple test so you know the proxy config is actually being used.
codex -m gpt-5.4 -c model_reasoning_effort="high" -c model_verbosity="medium"Use the desktop app if you prefer a dedicated Codex workspace.
{
"OPENAI_API_KEY": "paste-your-aimesh-api-key-here"
}Need a different model or reasoning profile? Use the selector page to generate a fresh config.toml and relaunch command.