Good general models
Many bad models shilled, going to make list of actually good models in my opinion from personal experience. Models will be removed/changed over time
Text Generation • Updated • 74.4k • 661Note Great at translation, otherwise surpassed by Llama 3.1
mattshumer/Reflection-Llama-3.1-70B
Text Generation • Updated • 37.1k • 1.65kNote Slow because of reflection training, but this also means it is smarter and punches above it's weight despite being 70B. Overall best 70B model if you don't mind waiting a lot, worse than base Llama 3.1 at translation I think?
meta-llama/Llama-3.1-70B-Instruct
Text Generation • Updated • 746k • • 539Note Best base 70B model currently and a good safe pick
mistralai/Mistral-Nemo-Instruct-2407
Text Generation • Updated • 83.1k • • 1.1kNote Nice mix between speed and smartness
mistralai/Mistral-Large-Instruct-2407
Text Generation • Updated • 19.5k • 746Note Seemingly close to Llama 3.1 405B but much smaller, I have only tested on Q3 and seems really good
google/gemma-2-27b-it
Text Generation • Updated • 167k • 409Note Really nice mix of speed and inteligence
google/gemma-2-9b-it
Text Generation • Updated • 729k • 473Note Also nice mix of speed and intelligence, but I would say if you can run Mistral-Nemo instead