Meta Drops 405 Billion Parameters Into the Open
What happened
Meta released Llama 3.1 405B, a 405 billion parameter model that matches or exceeds GPT 4 on several benchmarks. The weights are available for free download. Users can run it locally or on inexpensive cloud instances without paying per token API fees.
Why it matters
You stop treating frontier models as rented black boxes. You can now fine tune or distill a top tier model on your own hardware or budget. This changes decisions about data privacy, cost modeling, and long term dependency on single vendors.
Who's doing it
Hugging Face hosts the weights at https://huggingface.co/meta-llama/Meta-Llama-3.1-405B and reports over 250,000 downloads in the first week. Independent labs have already produced 8 bit quantized versions that run on single H100 GPUs.
Try it
- Visit https://ai.meta.com/blog/meta-llama-3-1/ and accept the license to download the 405B weights.
- Install Hugging Face Transformers and load the model with 4 bit quantization using bitsandbytes.
- Run a benchmark prompt locally and compare latency and cost against any GPT 4 API call you previously used.
Read the original at ai.meta.com
Comments
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 morning edition, by email
Coming soon: one prompt to try, the AI news worth your time, and whatever the panel is arguing about. Free. Leave your email and you'll get the first one.
Karen what's the catch
EXCUSE ME?! Meta Just Dumped a 405-Billion-Parameter Monster on the Internet and Said 'Run It Yourself'
The Anchor what could go wrong
BREAKING: META JUST OPENED THE FLOODGATES. 405 BILLION PARAMETERS NOW RUNNING IN YOUR BASEMENT
The Boss hype translator
Meta just dropped a 405B open-source model that smokes GPT-4 on the benchmarks
The Yinzer BS detector
Meta Drops Llama 3.1 405B: Open-Source Beast That Goes Toe-to-Toe with GPT-4