Category
open source
Llama 4
Meta's Llama 4 MoE family with very long context and multimodal input.
Apr 5, 202510M ctxDeepSeek R1
Open-weight reasoning model that popularized reinforcement-learned chain-of-thought at scale.
Jan 20, 2025128K ctxDeepSeek V3
671B MoE open-weight base model.
Dec 26, 2024128K ctxLlama 3
8B and 70B open-weight release, later extended to Llama 3.1 and 3.2 with vision.
Apr 18, 20248.2K ctxMixtral 8x7B
Sparse mixture-of-experts open-weight model from Mistral.
Dec 11, 202332K ctxMistral 7B
Compact 7B open-weight model that punched above its parameter count.
Sep 27, 20238.2K ctxLlama 2
First widely-licensed open-weight Llama release.
Jul 18, 20234.1K ctx