Sakana AI Launches Fugu, a Single API That Orchestrates Multiple AI Models
Sakana AI introduced Fugu, a multi-agent orchestration system that exposes one API while routing requests across a pool of large language models, or LLMs. The launch includes two versions delivered through an OpenAI-compatible interface: Fugu for lower-latency everyday tasks such as coding and chatbots, and Fugu Ultra for harder multi-step work including AI research, cybersecurity analysis and patent investigations.
Fugu is described as an LLM trained to call other models, including recursive instances of itself, and to handle model selection, delegation, verification and synthesis behind one endpoint. Sakana said Fugu Ultra matches the performance of Fable and Mythos on engineering, scientific and reasoning benchmarks and described the architecture as a way to route around vendor restrictions or export-control disruptions; the launch thread did not specify what share of the agent pool relies on open or closed models.
From the sources (8 posts)
@sakanaailabsIntroducing Sakana Fugu: A full multi-agent orchestration system accessible via a single model API. Our ‘Fugu Ultra’ model matches the performance of Fable and Mythos, delivering frontier capability without the risk of export controls. Tr
@hardmaruRT @SakanaAILabs: Introducing Sakana Fugu: A full multi-agent orchestration system accessible via a single model API. Our ‘Fugu Ultra’ mod…
@sakanaailabsFugu stands shoulder-to-shoulder with leading models like Fable and Mythos across the industry's most rigorous engineering, scientific, and reasoning benchmarks. Read the full blog: Beyond Bigger Models: Why are Or
@hardmaruRT @SakanaAILabs: Fugu stands shoulder-to-shoulder with leading models like Fable and Mythos across the industry's most rigorous engineerin…
@sakanaailabsHow does it work? Sakana Fugu is itself an LLM, trained to call various LLMs in an agent pool, including instances of itself recursively. Fugu dynamically orchestrates the world's best models to tackle complex, multi-step tasks. As shown
@hardmaruRT @SakanaAILabs: How does it work? Sakana Fugu is itself an LLM, trained to call various LLMs in an agent pool, including instances of it…
@eliebakouch> without the risk of export controls lol this is a closed source llm orchestrator that relies partly on closed source models api, they don't even report what percentage of closed vs open models the system uses to achieve these scores o
@matthewberman“Without the risk of export controls” 🔥🔥