9 stories
Only Kolibri's weights are out. Beam and Mistral Large 4 stay API previews until late October, and Kolibri's 4-bit build runs at 50 tok/s on a 64GB Mac.
Kolibri 1 is Aleph Alpha's 78B mixture-of-experts model for German and English, with 1M-token context and Apache 2.0 licensing.
Kolibri-1 is Aleph Alpha's 78B MoE model with 3.5B active params, 1M context, and Apache 2.0 license, built for German and English.
Seattle-based artificial intelligence research firm Allen Institute for AI announced a development framework for large language models Thursday that significantly improves how mixture-of-experts large language models are trained.
Architecture note on GLM-5.3-Flash, formerly Ox Alpha, covering its KDA and MLA/DSA hybrid attention, sparse MoE backbone, and four-stream mHC residual path.
No released vLLM build served Kimi K3 when we started. We resolved several distinct failures on 8x B300 and measured the result, its operating crossover, and its honest limits.
Short architecture note on Kimi K3, including LatentMoE, Kimi Delta Attention, Attention Residuals, NoPE, multimodality, and inference-efficiency choices.
Short note on the architectures of six new open-weight models, including Nanbeige 4.2, Laguna S 2.1, Motif-3-Beta, Solar Open 2, Antares 1B, and BTL-3.
Architecture and benchmark notes on Thinking Machines Lab's 975B Inkling MoE, including short convolutions, relative-position bias, training, and effort control.
Showing 9 of 9 stories
You're all caught up.
Get the top AI news and tools straight to your inbox every week.
Weekly · AI news
Join 15,000+ AI enthusiasts
