aiqbench analysis
Cursor Research's open-sourcing of Mixture-of-Kittens (MoK) represents a significant optimization for deterministic MoE training, achieving up to 2.37x throughput gains on NVIDIA Blackwell GB300 NVL72 racks by fusing communication and computation into a single kernel. However, the requirement for high-density GPU clusters effectively limits adoption to large-scale infrastructure providers or well-funded labs, excluding smaller research teams lacking access to such hardware floors. This development underscores the growing divergence between cutting-edge training efficiency gains and their accessibility within the broader open-source community.
Cursor Open-Sources Mixture-of-Kittens (MoK): A Deterministic MoE Training Megakernel for GB300 NVL72 Racks
Cursor Research has open-sourced Mixture-of-Kittens (MoK), the MoE training megakernel behind its Composer models. MoK fuses all mixture-of-experts communication and computation into a single deterministic kernel, and runs up to 2.37x faster than the strongest public baseline on GB300 NVL72 racks. It requires Blackwell SM100 or SM103 GPUs, which puts it out of reach for anyone without NVL72 capacity. The post Cursor Open-Sources Mixture-of-Kittens (MoK): A Deterministic MoE Training Megakernel for GB300 NVL72 Racks appeared first on MarkTechPost.
Cursor Research has open-sourced Mixture-of-Kittens (MoK), the mixture-of-experts training megakernel behind its Composer models. MoK fuses every MoE communication and computation step into a single deterministic kernel. Cursor team reports up to 2.37x higher throughput than the strongest public baseline. It already powers Composer training across tens of thousands of GPUs. Yes, but the hardware floor is high. MoK is on GitHub under Apache-2.0. It requires NVIDIA Blackwell SM100 or SM103 GPUs, which means GB200 NVL72 or…
Excerpt shown under the publisher's RSS terms · 77 words · full text and figures are on the publisher's site
Related coverage
[AINews] OpenAI reports Navier-Stokes singularity find in 88 hours using Astra-next, roughly 10,000 agents and 130B tokens (>$40M), a contender for second ever Millennium Prize awarded
Latent Space · Sep 9, 2026
[AINews] not much happened today
Latent Space · Sep 10, 2026
Mixture-of-Kittens: MoE Megakernel for NVL72s
arXiv · Sep 30, 2026
[AINews] Meta Muse Wonder & Wonder Full: MSL’s Claude 5.5-tier Frontier Model
Latent Space · Oct 13, 2026