Sakana AI's Fugu: New AI Outperforms Claude Fable 5

Japanese startup Sakana AI has launched a new AI system called Fugu. This system coordinates multiple language models through a single API, rather than relying on one foundation model. Sakana AI claims its Fugu Ultra version matches or surpasses Anthropic's Claude Fable 5 on select coding, science, and reasoning benchmarks. The company states Fugu Ultra scored 93.2 on the LiveCodeBench coding benchmark, compared to Claude Fable 5's 92.9. Fugu aims to offer greater resilience against AI vendor restrictions. This multi-model approach allows it to tackle complex tasks and reduces dependence on any single AI provider. Fugu splits user requests, sends them to specialized AI models, and then merges the answers. There are two versions: Fugu for everyday tasks, and Fugu Ultra for advanced workloads like AI research and cybersecurity. This new approach could change how businesses use AI by offering a more flexible and resilient solution.