Stories tagged “open weights” on AI Learning Ground Buzz.
DeepSeek published open weights for DeepSeek-V4-Pro (1.6T total, 49B active parameters) and V4-Flash (284B total, 13B active), both with 1M-token context and thinking and…
DeepSeek
Google DeepMind released Gemma 4 in four sizes (effective 2B and 4B edge models, a 26B mixture-of-experts and a 31B dense model) under the Apache 2.0 license. The models support…
Mistral Small 4 is a 119B-parameter mixture-of-experts model with about 6B active parameters per token, a 256k context window and image input, released under Apache 2.0. It…
Mistral AI
Alibaba's Qwen team released Qwen3.5-397B-A17B, a mixture-of-experts model with 17B active parameters, a 1M-token context window and native vision-language support, alongside a…
Alibaba Qwen
Z.ai launched GLM-5, a 744B-parameter mixture-of-experts model with about 40B active parameters, released with open weights under the MIT license. It targets complex systems…
Z.ai
DeepSeek released DeepSeek-V3.2 as its new production model and V3.2-Speciale as a high-compute variant aimed at hard reasoning problems. The release builds on DeepSeek Sparse…
DeepSeek
Moonshot AI released Kimi K2 Thinking, an open-weight trillion-parameter mixture-of-experts model with 32B active parameters, trained to interleave reasoning with tool calls…
Simon Willison
