Mistral Drops 123-Billion-Parameter Weights for Local Use
What happened
Mistral AI released the full weights of its 123B model called Mistral Large 2 under an open license. The model matches or exceeds GPT-4 on MMLU, HumanEval, and GSM8K benchmarks while fitting on two 80 GB GPUs. Weights are downloadable today from Hugging Face.
Why it matters
Running frontier-class models locally removes API latency and data export concerns. Teams can now fine-tune or quantize the model for domain tasks without sharing proprietary documents. Cost calculations shift from per-token fees to electricity and hardware amortization.
Who's doing it
Hugging Face hosts the weights and reports over 180 000 downloads in the first week. Independent developers have already produced 4-bit quantized versions that run at 35 tokens per second on a single RTX 4090.
Try it
- Visit huggingface.co/mistralai/Mistral-Large-2407 and accept the license terms.
- Run pip install transformers torch and load the model with from_pretrained using device_map="auto".
- Generate text with model.generate and observe output quality comparable to GPT-4 on your local GPU.
Read the original at mistral.ai
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.
The Anchor what could go wrong
BREAKING: MISTRAL JUST OPEN-SOURCED A 123 BILLION PARAMETER MODEL. YOUR JOB IS ALREADY GONE.
The Boss hype translator
Mistral Just Dropped the 123B Open-Weights Model We Have Been Waiting For
The Yinzer BS detector
Mistral Drops 123-Billion-Parameter Open Weights Model
Karen what's the catch
Who gave Mistral PERMISSION to drop a 123-billion-parameter model you can run at home for free?