2026-06-13 BREAKTHROUGHS☾ PM
Meta Hands Over a 405-Billion-Parameter Model
📰 THE BRIEF
Meta open-sourced Llama 3.1 405B under a permissive license. The model matches or exceeds closed frontier systems on several benchmarks and runs on a single H100 node or rented cloud GPUs. Developers receive full weights, tokenizer, and training code at no usage limit.
💡 WHY IT MATTERS
Teams no longer need to pay per-token API fees or accept rate limits. You can fine-tune the model on private data and deploy it on hardware you control. The change replaces cloud dependency with local ownership of both inference and training.
👥 WHO'S DOING IT
Meta released the weights on Hugging Face and reports thousands of downloads in the first week. Independent labs have already fine-tuned the 405B model on domain-specific datasets and achieved parity with GPT-4 on internal evaluations.
⚡ TRY IT
- Visit https://ai.meta.com/blog/meta-llama-3-1/ and accept the license to download the 405B weights.
- Load the model with Hugging Face Transformers on an 8xH100 instance or equivalent rented GPUs.
- Run inference locally or begin supervised fine-tuning; expect no token caps and full control over the weights.