# Sakana AI Releases Fugu Max and Fugu Ultra v2, Models That Orchestrate Other Models Instead of Being One URL: https://edufabtech.com/news/sakana-ai-fugu-max-ultra-v2-orchestration Section: News · Artificial Intelligence Published: 2026-09-13 · Updated: 2026-10-19 Publisher: EduFabTech (https://edufabtech.com) > Sakana AI's Fugu Max and Fugu Ultra v2 route tasks across pools of smaller, often open-weight models instead of relying on one large network, reporting leading agentic-benchmark scores without any proprietary frontier model in the loop — at a fraction of frontier per-token pricing. On 11 September 2026, Sakana AI announced two new releases, Fugu Max and Fugu Ultra v2, that do not add a bigger language model to the field so much as remove the assumption that one big model has to do all the work. Both are orchestrator systems: they take a task, break it up, and dispatch the pieces to a pool of smaller open-weight and specialist models, then assemble the results. Sakana AI, a Tokyo-based lab, describes the pair as "the same core orchestration architecture optimized for two distinct missions" — Fugu Max for cost efficiency, Fugu Ultra v2 for maximum output quality on hard, multi-step problems. ![](https://edufabtech.com/uploads/2026/09/sakana-ai-fugu-max-ultra-v2-orchestration-2.jpg) *Own work* ## What actually changed This is the latest iteration of an idea Sakana AI has been building since opening a beta of the original Fugu system in April 2026, ahead of a general release that June. That first version was pitched against a specific inefficiency: in the company's words, "conventional approaches to utilizing foundation models often require users to manage multiple API keys, as models from different providers tend to specialize in distinct areas." In the September release announcement, Sakana AI frames the same problem more bluntly, arguing that "a system that deploys a multi-trillion-parameter model to execute a simple data lookup is not intelligent, but wasteful." Instead of one model doing everything, a Fugu-trained router decides which of several available models — including, per the September release, models from NVIDIA's Nemotron line — should handle which part of the job, in what order, and whether the result needs a second pass. The architecture and training method behind this are documented in a technical report Sakana AI's research team posted to arXiv in June 2026, which describes combining large-scale fine-tuning, evolutionary search and reinforcement learning to teach a model to construct agent scaffolds on the fly rather than follow fixed routing rules. ## How the routing works In its April 2026 beta announcement, Sakana AI described the original Fugu system as one that "learns to dynamically assemble agents from a pool and coordinate them through non-obvious but highly efficient collaboration patterns," rather than following hand-written team structures. The same April announcement pointed to what it called "an interesting future direction" from its research: when a Fugu model is allowed to call itself recursively, "a new form of test-time scaling emerges," with the depth of recursion adjustable at inference time without retraining. The September announcement does not describe this as a feature of Fugu Max or Fugu Ultra v2. ## The benchmark numbers Sakana AI reported that Fugu Ultra v2 scored 48.3 on Chartography, a visual reasoning and data-interpretation benchmark, in its September 2026 release, compared with 27.3 for Anthropic's Opus 5 and 29.5 for its Fable 5 on the same test. On DeepSWE, a real-world software-engineering benchmark, Fugu Ultra v2 scored 74.3, which Sakana AI says places it ahead of models priced several times higher per token. Across the eight agentic benchmarks Sakana AI uses for evaluation, it reports Fugu Ultra v2 placing best or joint-best on five and in the top two on seven, while Fugu Max separately takes the best overall score on six benchmarks, including Terminal Bench 2.1, GPQA Diamond and AA-LCR. The company states plainly that "Fugu Ultra v2 achieves these scores without Fable 5, Fable 5.1, or GPT-6-Astra in its agent pool" — the point being that the orchestration layer, not access to any single proprietary frontier model, is producing the result. On pricing, Sakana AI's console lists Fugu Max at a flat $2 per million input tokens and $6 per million output tokens regardless of context length, while Fugu Ultra v2 costs $5 per million input tokens and $30 per million output tokens for context up to 272,000 tokens, rising to $10 and $45 per million respectively beyond that threshold. ## What an independent read adds An independent analysis published by Pondero the following day flagged two caveats worth keeping in view. First, Fugu Max's headline comparisons are against similarly priced models, not the current frontier leaders, so the efficiency claim is narrower than it first appears. Second, the strongest results cluster in specific domains — visual reasoning and coding — and Sakana AI's own release does not claim the same margin generalizes to other task types. Pondero's analysis also noted that Sakana AI's pricing above the 272,000-token threshold was, at the time of its writing, still unclear; that detail has since been published on Sakana AI's console pricing page, cited above. ![A bar chart shows Fugu Ultra v2 scoring 48.3 on the Chartography benchmark versus 29.5 for Fable 5 and 27.3 for Opus 5, alongside its 74.3 DeepSWE score and top-two finishes on 7 of 8 agentic benchmarks.](https://edufabtech.com/uploads/2026/10/sakana-ai-fugu-max-ultra-v2-orchestration-3.png) *A bar chart shows Fugu Ultra v2 scoring 48.3 on the Chartography benchmark versus 29.5 for Fable 5 and 27.3 for Opus 5, alongside its 74.3 DeepSWE score and top-two finishes on 7 of 8 agentic benchmarks. — EduFabTech · Own work* ## Why it matters For engineering teams building agentic pipelines, the significant claim here is not a benchmark score but the architectural bet: that coordinating several smaller, swappable, often open-weight models can match or beat a single proprietary frontier model on real tasks, at a fraction of the per-token cost, and without being locked to any one provider's roadmap. Whether that holds outside Sakana AI's own benchmark suite is something outside labs and independent evaluators will need to test directly, ideally by running the same tasks against Fugu Max and Fugu Ultra v2 under conditions the vendor did not choose. ## Sources 1. Sakana AI. (2026). Introducing Fugu Max and Fugu Ultra v2: Orchestrating the Pareto Frontier. Sakana AI https://sakana.ai/fugu-max-release/ 2. Sakana AI. (2026). Sakana Fugu: A Multi-Agent Orchestration System as a Foundation Model. Sakana AI https://sakana.ai/fugu-beta/ 3. Yujin Tang, Edoardo Cetin, Jinglue Xu, et al.. (2026). Sakana Fugu Technical Report. arXiv https://doi.org/10.48550/arXiv.2606.21228 4. Sakana AI. (2026). Fugu API Pricing. Sakana AI https://console.sakana.ai/pricing 5. Pondero. (2026). Sakana AI ships Fugu Max and Fugu Ultra v2, beating frontier benchmarks without frontier models. Pondero https://pondero.ai/news/2026-09-12-sakana-fugu-max-ultra-v2/ --- Source page: https://edufabtech.com/news/sakana-ai-fugu-max-ultra-v2-orchestration Quote with a link to the source page. © EduFabTech.