Anyone tested it at high context yet? I find all Mistral models peter out after like 16K-24K tokes no matter what they advertise the context length as.
this post was submitted on 17 Mar 2025
28 points (96.7% liked)
LocalLLaMA
2792 readers
5 users here now
Welcome to LocalLLaMA! Here we discuss running and developing machine learning models at home. Lets explore cutting edge open source neural network technology together.
Get support from the community! Ask questions, share prompts, discuss benchmarks, get hyped at the latest and greatest model releases! Enjoy talking about our awesome hobby.
As ambassadors of the self-hosting machine learning community, we strive to support each other and share our enthusiasm in a positive constructive way.
founded 2 years ago
MODERATORS
A GPT-4o-mini comparable system that you can run on a RTX 4090 isn't going to solve direct problems, but it might have enterprise uses. Text generation automation for personal use should be strong, for example - in place of having a third party API do it.
Is this expected to be released on ollama?
I'm using https://ollama.com/justinledwards/mistral-small-3.1-Q6_K
And it's stunningly good. Absolutely running circles around Gemma 3 and Phi-4
This is so exciting! Glad to see mistral at it with more bangers.