Mistral 7b

Mistral AI's original 7B instruct model — the first fine-tuned release from Mistral, Apache 2.0 licensed and widely deployed.

Mistral AI's improved 7B instruction model — stronger chat performance and longer context than v0.1.

Mistral AI's landmark 7B base model — outperforms Llama 2 13B on most benchmarks, Apache 2.0 licensed.