A single place
for your API access.
Connect your research, notebooks, and coding tools using your own lab API keys.
Welcome back
Sign in with your existing lab account.
Use the same email and password as the lab workspace. API access requires an approved account.
Your account is awaiting approval
Read the setup guide →API keys
A key for each tool. One shared account limit.
3 concurrent requests, across all your keys
Requests may wait up to 30 seconds for a slot. A full account then returns
429.
Your keys
| Name | Key | Last used | Actions |
|---|
Loading your keys…
Your key list will appear here.
Usage
API activity across all of your keys.
Loading usage…
Requests
—
Total tokens
—
Estimated API cost
—
Token breakdown
Recorded usage- Input tokens
- —
- Cached input tokens
- —
- Output tokens
- —
Cost is an estimate from recorded usage, not a subscription charge. Totals use the gateway’s recorded token counters. Missing cost data appears as unavailable. Completed requests usually appear within about 60 seconds; in-progress streams are not counted live.
Connect your tools
Keep your client. Change the endpoint and API key.
https://getapi.ionclocklab.com/v1
For the OpenAI SDK, include /v1 in the base URL. Authenticate with
your lab API key using Authorization: Bearer.
Use your own key in place of the placeholder.
https://getapi.ionclocklab.com
For the Anthropic SDK and Claude Code, use the root URL without /v1.
The client appends /v1/messages. This format serves the available lab
models.
Use your own lab API key, not a shared provider credential.
Models & reasoning
From this gatewaySign in to see the available model IDs.
Both formats use the same OpenAI model IDs and shared lab allowance. The API format selects the request protocol; an Anthropic-compatible request does not switch the underlying model to Claude.
- Responses
reasoning: {"effort": "medium"}- Chat Completions
reasoning_effort: "medium"
Start with medium effort. Support for other reasoning and thinking controls depends on the model and gateway.
Limits & troubleshooting
- 3 concurrent requests
-
The limit belongs to your account and is shared across every key and API
format. Requests can wait up to 30 seconds for a slot; a full account then
returns
429. 401· Authentication- Check that the key is active, copied completely, and sent to this gateway. A lab password is not an API key.
400· Request or model-
Use an exact model ID from
GET /v1/models. Check the request format and any model-specific parameters. 429· Capacity-
Reduce parallel calls and honor
Retry-Afterwhen present. Creating another key does not increase the account limit. - When usage appears
- Completed requests usually appear within about 60 seconds. Usage collection does not report live partial tokens from an in-progress stream.
- Using an SDK
- SDK retries can resend requests. Configure retries deliberately, especially after a network interruption. For streaming, consume or close the response.
Keep API keys in environment variables or a secret store. Do not put them in browser-side code or commit them to a repository.