Sakana AI's quality-optimized multi-agent orchestration model. Delivered as a single OpenAI-compatible model, Fugu Ultra coordinates a pool of frontier models — selecting, delegating, verifying, and synthesizing internally — for hard, multi-step reasoning.
OffRail routes requests to the best providers that are able to handle your prompt size and parameters.
Sakana AI's quality-optimized multi-agent orchestration model. Delivered as a single OpenAI-compatible model, Fugu Ultra coordinates a pool of frontier models — selecting, delegating, verifying, and synthesizing internally — for hard, multi-step reasoning. You can access it through OffRail's OpenAI-compatible API with automatic provider routing, fallback, and cost analytics.
Pricing for Fugu Ultra on OffRail starts at $5.00 per million input tokens and $30.00 per million output tokens, depending on the provider. The pricing table above always reflects the current per-provider rates.
Fugu Ultra supports a context window of up to 1,000,000 tokens on its largest provider deployment.
Fugu Ultra is served by Sakana AI through OffRail. Requests are automatically routed to the best available provider, with fallback when a provider has issues.
Fugu Ultra supports tool (function) calling and soft JSON output, but NOT strict JSON output schema (its upstream provider does not enforce it).
Fugu Ultra was released on June 22, 2026.