Fugu Max (Sakana AI)

Fugu Max is a Sakana AI model that became available on September 11, 2026, landing alongside the Fugu Ultra v2.0 release in the same Sakana AI wave that month. Third-party cataloging describes it as a cost-optimized orchestration model designed to dynamically route tasks across open and specialized backends, aiming to improve cost-performance trade-offs rather than relying on a single monolithic architecture. That routing-first positioning suggests a practical fit for teams that want Sakana AI's coordination layer to pick the most suitable underlying model for each request, balancing capability against per-token spend.

From a deployment perspective, Fugu Max is delivered as a cloud-hosted API, and third-party cataloging reports a standard pricing tier that reflects its economy-oriented positioning. The catalog evaluation describes its behavior as frontier-tier in reasoning and coding, built on a robust autoregressive transformer with standard attention patterns that produce predictable, coherent token generation. For practitioners, this points to a sensible match for enterprise pipelines where consistent output quality matters more than peak specialization, particularly for reasoning-heavy workflows, code generation, and general text generation tasks where per-token economics are a primary concern.