Quick answer
OpenRouter documents routed model context as Varies by model; exposed as context_length. OpenRouter model metadata also exposes provider-specific per-request limits when available.
Verified Aug 22, 2026Official source
Current limits
| Constraint | Current value | Scope | Verified source |
|---|---|---|---|
| routed model contextOpenRouter model metadata also exposes provider-specific per-request limits when available. | Varies by model; exposed as context_length | Per model | OpenRouterAug 22, 2026 |
Why does this limit matter?
Changing the routed model can change context and output ceilings without changing the API interface.
This value is scoped to Per model; a different plan, runtime, model, endpoint, region, or account can produce a different effective constraint.
What should you check?
- Fetch the model object and record context_length and per_request_limits before dispatch.
- Confirm the exact plan, model, runtime, endpoint, region, and account that serve the failing workload.
- Record the observed value, response headers or configuration, timestamp, and source without logging secrets.
Important caveats
- Provider routes can differ, so the effective limit must be read from current metadata for the chosen route.
- Treat the official source and live account configuration as authoritative if they differ from this verified snapshot.
HyperObserve reports the documented platform constraint. Your application, SDK, gateway, provider, region, or account can impose a lower effective limit.
Related