Mixtral 8x7B Instruct is a pretrained generative Sparse Mixture of Experts, by Mistral AI, for chat and instruction use. Incorporates 8 experts (feed-forward networks) for a total of 47 billion parameters.
Instruct model fine-tuned by Mistral. #moe
Providers for Mixtral 8x7B Instruct (nitro)
With Nitro, requests are routed to the provider with the highest throughput, with fallbacks to maximize uptime.