Llama 3.1 405B: open-source reaches the frontier
In one sentence Meta releases Llama 3.1 405B under commercial license: for the first time an open model directly competes with GPT-4 and Claude 3.5 Sonnet on benchmarks, with 128K context.
Meta releases Llama 3.1 in three sizes: 8B, 70B, and for the first time 405B (four hundred five billion parameters). It's the largest open model ever released, and stated benchmarks put it on par with GPT-4 and Claude 3.5 Sonnet.
All three models now have a 128,000-token context window (was 8K in 3.0), 8-language support, native function calling.
Mark Zuckerberg publishes an open letter — "Open Source AI Is The Path Forward" — which becomes the manifesto of Meta's strategy. The release reframes the conversation: "open vs closed" is no longer a quality choice, it's a lock-in choice.
Companies
Meta
Tools
Llama 3.1
Tags
Sources