This AI Uses Other AIs to Solve Problems

TL;DR: Sakana AI's new Fugu Ultra model is now on Vercel's AI Gateway. Instead of a single model, it acts as a coordinator, routing tasks to a team of other AIs and combining their answers into one.
Key facts
- Category
- AI
- Impact
- High
- Published
- Source
- Vercel Blog
Full summary
Sakana AI's new Fugu Ultra model coordinates a team of other AIs to solve problems and is now available on Vercel.
Sakana AI's new model, Fugu Ultra, is now available for developers through Vercel's AI Gateway. Fugu Ultra operates differently from traditional large language models. Instead of being a single, massive model, it functions as a coordinator for a pool of other publicly available AI models. When it receives a prompt, it intelligently routes the task to a team of one to three specialized "agents," which are other AI models best suited for the job. It then combines their individual outputs into a single, coherent answer. According to reasoning and scientific benchmarks, this multi-agent approach allows Fugu Ultra to achieve performance comparable to other leading frontier models.
This integration matters for developers and CTOs because it makes a sophisticated multi-agent architecture easily accessible. By using Vercel's AI Gateway, developers can leverage Fugu Ultra's capabilities through a unified API without the complexity of managing multiple model integrations themselves. The gateway handles the intricate routing, performance monitoring, and cost management, simplifying the development process. This allows teams to build more powerful and nuanced applications by tapping into the collective intelligence of several models, each with its own strengths. It represents a practical way to implement advanced AI strategies without needing a dedicated research team.
The release of Fugu Ultra on a major platform highlights a significant trend in the AI industry. Companies are increasingly exploring ways to combine multiple smaller, specialized models rather than relying solely on scaling up single, monolithic ones. This approach can lead to greater efficiency, better performance on specific tasks, and more resilient systems. Making these complex architectures available through simple APIs lowers the barrier to entry for developers everywhere, potentially accelerating the next wave of innovation in AI-powered software and services. It signals a move toward more dynamic and collaborative AI systems.
Related on Notifire
Related stories
Primary source: Vercel Blog