huggingface.co · Open Weight Models
Open weight 2.8T parameter mixture of experts model from Moonshot AI with 104B active parameters, native MXFP4 weights, image input and a 1M token context. K3.1 Pro is its post trained successor.
Kimi K3 (Moonshot AI) is one of 30 open weight model tools on Axiomi. See all open weight model tools or how we check facts.
// guides
Can you run Kimi K3 or K3.1 Pro locally? GGUF sizes from 594 GB to 1.56 TB, real memory needs, llama.cpp and vLLM setups, license limits and cheaper options.
// related
GLM (Z.ai)huggingface.coOpen weight model family from Z.ai under the MIT license. GLM 5.3 Flash is a 320B mixture of experts model with 18B active parameters, image and video input and a 1M token context.Open Weight ModelsOpen source★ 7.5kQwen3.8 Flash Nexthuggingface.coOpen weight multimodal MoE preview from Qwen (125B total, 6B active, plus a 51B n gram embedding table) under the Qwen Community License. Runs locally via Unsloth GGUF, ISTA GSQ builds, Ollama MLX, Strata, vLLM and SGLang.Open Weight ModelsOpen sourceInkling 2 Small (Thinking Machines)huggingface.coOpen weight multimodal mixture of experts model from Thinking Machines Lab under Apache 2.0, with 594B total and 17B active parameters, native NVFP4 weights, image input and a 1M token context.Open Weight ModelsOpen source
Mistral AImistral.aiParis based lab publishing both open weight and commercial models, from small edge models to mixture of experts flagships.Open Weight ModelsFreemium★ 10.8k
DeepSeek-V3github.comLarge mixture of experts model from DeepSeek whose technical report detailed efficient training tricks such as multihead latent attention.Open Weight ModelsOpen source★ 104.8k
Llamallama.comMeta's family of open weight language models, released with weights for research and commercial use under the Llama community licence.Open Weight ModelsSource available★ 7.7k