OpenAI-compatible client

Prerequisites

• Install the published @unlimgpt/cli package.
• Setup signs you in and creates an API key if needed; the secret stays in the operating-system credential store.

Setup

• Run unlimgpt configure generic-openai to prepare the profile and key.
• Start your chosen program through unlimgpt launch openai: after --, supply its installed executable name and normal arguments. Only the child process receives the Base URL and key.
• If you configured a separate profile such as work, add --profile work before -- when launching; this uses that profile's key without changing the active profile.

Verify

• Send “Reply with READY” in the launched OpenAI-compatible client, wait for a non-empty response, and confirm the request in API Activity. The wording may vary; the request uses credits.
• If you configured --profile work, add the same flag before -- in the runnable example below, using your actual profile name; otherwise it uses the active profile's key.

This launch check requires Python 3 and makes one billable request through the child process environment. If gpt-5.6-sol is unavailable to your key, use an accessible model ID from the catalog; the secret does not appear in command arguments.

unlimgpt launch openai -- python3 - << 'PY'
import json
import os
import urllib.request

body = {"model": "gpt-5.6-sol", "input": "Reply with READY", "max_output_tokens": 64, "stream": False}
request = urllib.request.Request(
    os.environ["OPENAI_BASE_URL"].rstrip("/") + "/responses",
    data=json.dumps(body).encode("utf-8"),
    headers={"Authorization": "Bearer " + os.environ["OPENAI_API_KEY"], "Content-Type": "application/json"},
    method="POST",
)
with urllib.request.urlopen(request, timeout=60) as response:
    print(response.status, response.headers.get("request-id"))
    print(response.read().decode("utf-8"))
PY

Rotate

• Record the old key ID from unlimgpt keys list, then create a new key with unlimgpt keys create --name rotated-client.
• Relaunch the client through UnlimGPT, send a request, and confirm it in API Activity. Revoke the old key only afterward.

Revoke

• Revoke a key on the API keys page after confirming the new connection; the old key immediately loses access.

Revert

• No client files were changed: stop using the UnlimGPT launcher and revoke the key on the API keys page if it is no longer needed.

Troubleshooting

If the request fails, run unlimgpt doctor --client generic-openai and see troubleshooting. Diagnostics do not replace a real request.

Uninstall

• Stop using the launcher and revoke the dedicated key; no configuration file was created.