ModelFugu Max by Sakana AI
Text+image->text
Free credits included. No credit card required.
Fugu Max is the cost-performance model in Sakana AI's Fugu family, a learned multi-agent orchestration system that routes tasks across a large pool of open-weights and specialized models — including NVIDIA Nemotron models — to deliver frontier-grade results at a fraction of the token spend. Rather than a single monolithic model, Fugu dynamically assigns each subtask to the leanest capable agent, expanding the Pareto efficiency frontier between cost and capability. At $2/M input and $6/M output tokens, Fugu Max targets the best possible output per dollar among frontier-class systems, achieving top scores on benchmarks such as TerminalBench, GPQAD, and SWEFish.
