Back to Llama 3.1
Meta's 70B Llama 3.1 instruct model — the sweet spot of the 3.1 family, with 128K context and near-GPT-4-level performance.
131K tokensFree / Open weightsTransformerLlama 3.1 Community
Benchmarks
Llama-3.1-70B-Instruct
81.7%Llama-3.1-405B
49.8%AWorld (inclusionAI, GPT-4o + DeepSeek V3 + Claude + Gemini)
