Meta Drops the 405-Billion-Parameter Llama 3.1 With Full Commercial Rights
Meta published the full weights of Llama 3.1 405B under a commercial license that permits fine-tuning, distillation, and resale of derivative products. The release includes the model card, tokenizer, and reference implementations on Hugging Face and GitHub. No usage caps or revenue share requirements apply.
Teams no longer need to route frontier-scale inference through closed APIs or accept restrictive terms. You can now run or host the model locally, fine-tune domain data, and ship products without per-token billing. This changes cost modeling from variable API spend to fixed infrastructure spend.
Meta published the weights and license at ai.meta.com/blog/meta-llama-3-1. Within 48 hours, independent hosts such as Together AI and Fireworks offered 405B inference at roughly one-third the price of equivalent closed models.
Step 1: Visit huggingface.co/meta-llama/Meta-Llama-3.1-405B-Instruct and accept the license agreement. Step 2: Download the weights or point vLLM or Hugging Face Text Generation Inference at the repo to spin up a local endpoint. Step 3: Run a benchmark prompt to confirm output parity, then begin domain fine-tuning with your own dataset.