Quick answer
Grok documents grok 4.6 context as 500,000 tokens. Requests above 200K context have separate documented pricing.
Verified Aug 22, 2026Official source
Current limits
| Constraint | Current value | Scope | Verified source |
|---|---|---|---|
| Grok 4.6 contextRequests above 200K context have separate documented pricing. | 500,000 tokens | xAI API model grok-4.6 | xAIAug 22, 2026 |
Why does this limit matter?
Context capacity and higher-context pricing are separate planning dimensions.
This value is scoped to xAI API model grok-4.6; a different plan, runtime, model, endpoint, region, or account can produce a different effective constraint.
What should you check?
- Confirm the exact model ID and count prompt, media, reasoning, and reserved output tokens.
- 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
- Other Grok model IDs can expose different context and rate limits.
- 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