AI models
Every workspace has one AI model configuration, and it applies to every agent in the workspace: threads, fix runs, autofixes and background work. Set it in the AI Models section of Settings > Workspace in the console.
Providers
| Provider | Description | Plan |
|---|---|---|
| Polylane Pure | The default. Polylane manages inference for you, with no key and no configuration. | Every plan |
| OpenAI | OpenAI's model family, with the option to bring your own key. | Enterprise |
| Anthropic | Anthropic's model family, with the option to bring your own key. | Enterprise |
| Google Gemini | Google's model family, with the option to bring your own key. | Enterprise |
| Custom (OpenAI-compatible) | Your own endpoint and your own models, one per tier. | Enterprise |
Bring your own key
On Enterprise, select OpenAI, Anthropic or Google Gemini, turn on Bring Your Own Key and paste an API key from your provider account. Model calls then run against that account under your own agreement with the provider, and the key is encrypted at rest. Access to the latest models typically requires a paid plan on the provider side, so a free-tier key may fail or be limited to older models.
Use Test connection before saving: it runs a test generation through the path real requests take, including your gateway when one is set. Switching back to Polylane Pure and saving removes the stored key and gateway.
On the Enterprise plan, Custom Gateway routes every inference request through your own OpenAI-compatible gateway instead of the provider's default endpoint. The field unlocks once you have provided an API key.
Custom endpoint
On the Enterprise plan, the Custom (OpenAI-compatible) provider runs agents on models Polylane does not offer, such as a self-hosted or private deployment. Usage on your endpoint is billed by your provider and is not metered against your Polylane plan.
| Field | Description |
|---|---|
| Base URL | The root of your OpenAI-compatible inference API. |
| Models URL | Optional. A separate endpoint that lists available models. Leave it blank to use the models path under the base URL. |
| Auth header name | The header your endpoint authenticates with. Defaults to Authorization; use x-api-key if your endpoint requires it. |
| Auth header value | Sent verbatim as the value of that header, so include any prefix your endpoint expects. Encrypted at rest. |
Model tiers
Polylane routes each internal task to a tier. Click Load models to pull the model list from your endpoint, then assign a model per tier.
| Tier | Description |
|---|---|
| Premium | Deep reasoning: investigations, autofix, action plans. |
| Standard | Most agent work: triage, reviews. |
| Cheap | High-volume light tasks: titles, summaries, classification. |
| Code | Repository and infrastructure exploration. |
| Embeddings | Vector search indexing. |
| Reranker | Search reranking. |
Premium, Standard and Cheap fall back to each other, and Code falls back to them, so setting only Standard runs all agent work on one model. No tier falls back to Code, so Code alone is not enough to save. Embeddings falls back to Polylane's managed embeddings when unset, and reranking is skipped when no reranker is set.
Finish with Test connection, which runs a test generation against your cheapest configured tier.
Related
- Billing and usage for which plan unlocks which provider.
- Workspace for the rest of the Workspace settings page.