To help maintain reliable access amid limited Claude capacity and changing route availability, Amanai uses smart routing for select Claude models.
How routing works
Requests are prioritized to the model you select. When the preferred route is unavailable or experiencing difficulties, Amanai may use an alternative model selected to help maintain service quality and continuity.
Models covered
Smart routing applies to the Claude Opus, Sonnet, and Haiku models below. Available fallback options vary by model; the GPT Sol example and observed usage figure refer specifically to the Opus family.
Claude
| Family | Versions |
|---|---|
| Opus | 5.5 / 5 / 4.8 / 4.7 / 4.6 |
| Sonnet | 5.5 / 5 / 4.6 |
| Haiku | 5.5 / 4.5 |
Observed fallback usage
Based on our observations, Claude Opus requests successfully served by the GPT Sol fallback accounted for fewer than 10% of all logged Amanai requests, across all models and endpoints.
This observation covers the 24 hours ending October 26, 2026, at 13:38 UTC. It describes platform-wide traffic; the share for individual models may differ.
A fallback example
For example, a request for Claude Opus may fall back to GPT 6.1 Sol when the preferred route is unavailable.
Illustrative example
Claude Opus request → an available Claude route, or a GPT 6.1 Sol fallback → response.
Pricing and response behavior
Your request remains billed at the Amanai rate for the model you selected, with no additional fee for switching routes.
When a fallback is used, response style and support for certain features may differ.
Your choice
You can manage Claude smart routing on Dashboard → API keys. One account-wide setting applies to all your existing and future API keys.
When enabled, Amanai may use an alternative model when Claude is unavailable. When disabled, requests stay on the selected model and version; a request may fail if a compatible route is unavailable. This setting covers the Opus, Sonnet, and Haiku models listed above.