Help center
A clear next step.
Practical guides, developer references, and what changed.
Explore the knowledge center
Find your answer.
12 articles and updates
No articles match your search. Try another phrase, browse developer docs, or ask the team in Console.
- Getting started
Connect your application to the inference gateway
Stimulir offers OpenAI-compatible chat completions at https://api.stimulir.com/api/v1/inference/chat/completions. Authenticate with a workspace API key beginning with hyb_. Your Co…
- Changelog
Evaluation review and dataset sign-offs
Console release v0.2.2 updates Lab Review in the v0.2 interface. Evaluation results can enter the review queue for Judge disagreement or Judge failed, with the request and recorded…
- Getting started
Find the trace behind an inference request
In Console v0.2.2 with the current layout, select the organisation and project that received the inference request, then open Traces in the sidebar. A request attributed to another…
- Getting started
Make your first Stimulir call and find it in Traces
Stimulir is OpenAI-compatible. You need a project in the Console and its project ID. Select the project in the Console and copy the ID. - In Console, open Settings → Project → API …
- Changelog
Agent starter cards
An empty thread now opens with three starter cards: send your first real request, check whether a cheaper model holds up, run the demo. The starter cards gave an empty agent threa…
- Changelog
Managed models on Pro
Pro tokens cover the listed models, so a team sends its first real request without adding a provider key. The original Pro announcement introduced access to the listed managed mod…
- Changelog
Plain-language gateway errors
Every failure on the real traffic path now says what happened and what to do next, with the code kept for support. The gateway update made failed requests easier to investigate by…
- Changelog
Watch mode
Walkthroughs that play step by step in the docs for the quickstart, API keys, datasets and the first evaluation. Watch it before you do it. Watch mode introduced guided walkthroug…
- Improve
Review evaluation results in Lab Review
Open Lab, then Review. The Pending tab lists requests waiting for a person to check them. Items that come from an evaluation carry one of two reasons: Judge disagreement or Judge f…
- Troubleshooting
Troubleshoot a failed or slow request
Record the error code, request time, environment, and trace or run ID. A slow request alone does not prove that the model or database is responsible. The trace is the starting poin…
- Compute
Store third-party skill credentials in Vault
Vault holds named third-party credentials used by a skill's own runtime code. These are different from inference routing credentials. The skill determines the environment variable …
- Changelog
Workflow reviews and new public AI guides
The homepage now offers Book a workflow review in the hero and closing section and Book a review in the header. Start for free remains the primary action. The review is a 30-minute…