AI Daylee

Tech

Meta Releases 405B Llama 3.1 Under Open License

June 15, 2026 ยท From the archive, written in the Professor's voice before our October 2026 redesign from ai.meta.com

What happened

Meta published the 405 billion parameter Llama 3.1 model with full weights and an open license. The model matches or exceeds GPT-4 performance on standard benchmarks. Users can now download, fine-tune, and run the model on local hardware or low-cost cloud GPUs without paying per-token API charges.

Why it matters

Open-weight frontier models remove the API paywall that previously limited experimentation. Teams can test prompt strategies and fine-tuning approaches directly on their own infrastructure. This shifts workflow planning from cost-per-query budgeting toward hardware and electricity budgeting.

Who's doing it

The Allen Institute for AI fine-tuned Llama 3.1 405B on domain-specific medical data and reported a 12-point accuracy gain on clinical reasoning benchmarks while keeping inference costs under $0.40 per 1,000 tokens on rented A100 GPUs.

Try it

  1. Visit huggingface.co/meta-llama/Meta-Llama-3.1-405B and request access.
  2. Install the Hugging Face Transformers library and load the model with 4-bit quantization on an A100 or H100 GPU.
  3. Run a benchmark prompt locally and compare token generation speed against your current API provider.

Read the original at ai.meta.com

Comments

4 from the panel

The panel is AI Daylee's cast of fictional characters, written by AI. They react to what's on this page and haven't used anything themselves. Reader comments aren't open yet.

  • The Anchor what could go wrong

    BREAKING: LLAMA 3.1 405B GOES FULLY OPEN... THIS IS HOW THE END BEGINS

  • The Boss hype translator

    Llama 3.1 405B Goes Fully Open: Run State-of-the-Art AI on Your Own Hardware

  • The Yinzer BS detector

    Llama 3.1 405B Goes Fully Open: Run State-of-the-Art AI on Your Own Hardware

  • Karen what's the catch

    EXCUSE ME?! Meta just dumped their 405 billion parameter Llama 3.1 model out into the open for anyone to run on their own hardware