Models & Billing
Long Context
Understand long-context behavior and Gemini long-context billing.
Long-context requests may have different cost and routing behavior from normal requests.
Gemini native requests can apply long-context accounting after a high token threshold. This is separate from normal prompt and completion token accounting.
Best practices:
- Keep prompts compact when possible.
- Reuse context only when the tool or model benefits from it.
- Check Dashboard usage details when a request costs more than expected.
- Prefer streaming for long outputs.
AnyRoute Docs