This model is over 2 years old. The same company has since released something newer: Muse Glimmer 30B (2026-08-10). The details below are still accurate for this model — it just is not what we would recommend today.

Llama 3.1 405B

Meta AI🇺🇸 United States
deprecated

Meta's largest open-weights model. 405B parameters, instruction-tuned.

Context window128K tokens

Version History

Meta-Llama-3.1-405B-Instructmajor

Meta opens Llama 3.1 405B — the first open-weights model to challenge GPT-4o. Released with 128K context and tool use support.

Benchmark Scores

Full leaderboard →
26.0 tokens_per_sec
Speed (tok/s)