Documentation Index
Fetch the complete documentation index at: https://docs.guardway.ai/llms.txt
Use this file to discover all available pages before exploring further.
What this is for
Guardway keeps a per-gateway model inventory. Providers are scoped to a single gateway, so each(gateway, provider) pair tracks its own model list — useful when the same upstream service is configured on multiple gateways with different regions, entitlements, or local builds.
Use Configuration → Models to discover what each provider currently exposes on a given gateway and to toggle individual models on or off for routing and the playground.
Options
Sync behavior
- Manual sync — click Sync Models on a
(gateway, provider)row. The gateway calls the provider’s model-listing endpoint (or uses a static catalog for providers that don’t expose one), parses the result, and reports the inventory back to the dashboard. - Automatic sync — runs on gateway heartbeat. Manual sync is mostly for “I just rotated the key, pick up the new models now”.
Columns
| Column | Meaning |
|---|---|
| Name | Provider’s model id (e.g. gpt-4o-mini, claude-sonnet-4-5). |
| Gateway | Which gateway this inventory row belongs to. |
| Provider | Display name of the provider. |
| Type | chat, completion, embedding, image, or audio. |
| Context | Max context window reported by the provider. |
| Input / Output price | Per-million-token pricing if known, otherwise blank. |
| Requests / Latency | Rolling stats for this model on this gateway. |
| Enabled | Toggle. Disabled models are hidden from routing and the Playground. |
How to configure
Open Configuration → Models
Open Configuration → Models from the dashboard sidebar. Each row groups by
(gateway, provider).Sync the inventory
Click Sync Models on the row(s) you want to refresh. Wait for the toast confirmation; the inventory populates within a few seconds.
Enable or disable models
Use the Enabled toggle on each model. Disabled models still appear in the table but are blocked at routing time and filtered out of the Playground.
Verify in Playground
Open Playground and confirm the enabled models are selectable. Send a test request to confirm the gateway can route to the model.
Related
- Providers — attach the provider before models can be synced.
- Routing — choose how requests fan out across enabled models.
- Playground — test an enabled model end-to-end.