Llama 3.1 405B

Llama 3.1 405B is the most powerful open-weight model ever released, matching GPT-4 on multiple benchmarks while supporting a 128K-token context and a permissive license that lets thousands of companies and researchers fine-tune and deploy it without API fees. It achieves 89.5% on MMLU and offers inference costs at $0.70 per million tokens—70% cheaper than #1 GPT-4o and 60% cheaper than #3 Gemini Ultra 1.5, democratizing frontier AI for broader access.