Sakana AI, a Tokyo-based AI company, has released four products in rapid succession since March 2026, including Sakana Chat, Sakana Marlin, Sakana Fugu, and Sakana Translate. Sota Omura, head of produ…
Researchers introduced Dream-Cubed, a dataset of over 30 billion blocks from Minecraft, and trained discrete and continuous 3D diffusion models for biome-conditioned chunk generation, achieving high-f…
Researchers from New York University have released Dream-Cubed, a large-scale dataset of Minecraft worlds comprising tens of billions of cubes, and trained a family of transformers to generate fully p…
Sakana AI announced Fugu-Ultra v1.1, an upgraded orchestration engine that delivers performance gains of up to 7.9 points over v1.0 and beats leading models like Fable 5 in complex coding and reasonin…
Researchers at Sakana AI introduced UnMaskFork (UMF), a test-time scaling method that enables multiple masked diffusion language models to collaborate on a single answer, outperforming existing method…
Sakana AI released Fugu-Cyber, a new orchestration model for cybersecurity that achieves state-of-the-art performance with an 86.9% success rate on CyberGym and 72.1% on CTI-REALM, comparable to leadi…
Sakana AI is teaming with NVIDIA to integrate NVIDIA's open model stack, including Nemotron, into Sakana Fugu, Sakana AI's multi-agent orchestration system, aiming to advance open model innovation fro…
Researchers from IT University of Copenhagen, Sakana AI, and Autodesk have developed a system of hundreds of simple cubic bricks that each run identical local neural networks and communicate only with…
Researchers from MIT, NYU, and Sakana AI replicated the Picbreeder experiment using vision-language model agents, finding that while AI can generate diverse images without predefined goals, it struggl…
Researchers introduced Sheaf-ADMM, a framework for multi-agent coordination that divides complex tasks into overlapping pieces assigned to individual agents. In tests, it achieved a 93% solve rate on …
Researchers at ICML 2026 presented a unified framework showing that parametric and nonparametric black-box optimizers are variations of a single update equation. By bridging this theoretical gap, they…
An AI agent using the AutoResearch framework autonomously improved a small GPT's training recipe over 123 experiments on a single H100 GPU, achieving a best mean bits-per-byte (BPB) of 0.9774 ± 0.0019…
Sakana AI released Fugu Ultra, a multi-agent orchestration system that matches frontier models like Anthropic's Fable 5 and Mythos Preview on benchmarks while avoiding single-vendor dependency. The mo…
Sakana AI launched its first commercial product, Sakana Marlin, an autonomous research assistant for business that conducts up to eight hours of independent research and generates structured summary s…
Sakana AI has formally established the Recursive Self-Improvement (RSI) Lab, a dedicated research group tasked with developing autonomous AI systems that can improve themselves through continuous, evo…
Researchers have developed DiffusionBlocks, a method that trains neural networks by dividing them into blocks and training only one block at a time, reducing memory requirements to roughly 1/B of stan…
Researchers from NVIDIA have developed a new sparse data format and custom GPU kernels, called TwELL, that reshape unstructured sparsity in transformer language models to align with GPU architecture, …