MacWhisper
Use Nebul models for MacWhisper's AI features with a custom OpenAI-compatible service.
MacWhisper is a macOS transcription app. Transcription runs on your Mac, and its AI features (summaries, questions about a transcript, rewrites, custom prompts) run on an AI service you pick in the app. The Custom service accepts any OpenAI-compatible chat-completions endpoint, including Nebul.
Prerequisites
- A Mac running MacWhisper
- A Nebul AI Studio account with API access
- An API key for your active project
- A model ID from the Model Catalog
Connect it to Nebul
- Open Settings in MacWhisper and select AI Services.
- Choose Custom under All Services, or use the Add Service button.
- Enter the connection details:
| Field | Value |
|---|---|
| Name | Nebul |
| Base URL | https://api.inference.nebul.io/v1 |
| API Key | Your Nebul API key from AI Studio |
| Model ID | A model ID from the Model Catalog, for example zai-org/GLM-5.3 |
- Save the service.
The model ID is case-sensitive and must match the catalog exactly, vendor prefix included. Only models enabled for your project work.
Transcription is unaffected
AI services apply to the AI features, not to transcription. Transcription models are configured separately in the transcription picker, and local transcription stays local.
Run it
Open a transcription, then pick an AI feature such as a summary or the prompt box. Choose Nebul in the provider picker if it is not already selected. The response comes back from the model you configured, and usage appears under the API key's project in AI Studio.
Troubleshooting
- API key rejected: copy the complete key without spaces or quotes, and confirm it belongs to the Nebul project you think it does.
- Model cannot be found: use the exact model ID from the Model Catalog, including the vendor prefix.
- Requests fail on long transcripts: large transcripts can exceed a model's context window. Use a model with a larger context, or shorten the transcript or prompt.
Data and telemetry
MacWhisper processes your audio on your Mac. When you use an AI feature, it sends the transcript text and your prompt to the configured provider, which after this setup is Nebul rather than a third-party model API. MacWhisper documents this in Setting up AI providers.
For how Nebul treats the requests it receives, see Privacy & Security.