Lab release history
Last updated Apr 5, 2025
Meta AI model releases
Ships the open-weight Llama family. This page collects the lab's model releases, lifecycle events, source links, and model metadata in one crawlable record.
10 models
Llama 4 Maverick
AvailableMeta's flagship open-weight MoE; highest MMLU among open models at release.
Llama 4 Scout
AvailableEfficient open-weight MoE designed for very long context on modest hardware.
Llama 3.3 70B
AvailableLate-2024 70B Llama update delivering much of the 405B instruction-following quality at lower serving cost.
Llama 3.2 90B Vision
AvailableFirst Llama family release with native vision models, alongside smaller edge-oriented 1B and 3B text models.
Llama 3.1 405B
AvailableMeta's first frontier-scale open Llama model, with 405B parameters, 128K context, multilingual support, and tool-use improvements.
Llama 3 70B
AvailableFirst Llama 3 release, with 8B and 70B open models and a stronger tokenizer, data mix, and post-training stack.
Code Llama 34B
AvailableMeta's first code-specialized Llama model family, released in base, Python, and instruction-tuned variants.
Llama 2 70B
AvailableThe release that made capable open-weight models genuinely usable for production.
LLaMA
AvailableMeta's first LLaMA, released to researchers; its leak catalyzed the open-weight movement.
Galactica
WithdrawnA science-focused model whose public demo was withdrawn after just three days over confidently wrong outputs — an early, instructive retraction.