Mistral AI · Open-Weight Models

Mixtral 8x7B

Name: Mixtral 8x7B
Author: Mistral AI

Mistral AI's sparse mixture-of-experts model: eight experts, two active per token, matching much larger dense models at launch with lower inference cost.

Official page

Run it locally

With Ollama installed, download and start the default version with:

ollama run mixtral:8x7b

Variant	Download
mixtral:8x7b	26GB
mixtral:8x22b	80GB

Memory: About 26GB for the default mixtral:8x7b version (4-bit)
Formats: Safetensors (Hugging Face), GGUF (Ollama, llama.cpp, LM Studio)

The download size is roughly the memory (RAM or GPU VRAM) needed to load a variant; allow extra for long contexts. See our hardware guide.

Overview

Mixtral 8x7B, released in December 2023, popularized sparse mixture-of-experts among open models. Each token is routed to two of eight expert networks.

Highlights

46.7B total, about 12.9B active parameters
32K-token context window
Apache 2.0 license
A larger 8x22B version followed in 2024

Typical uses

Self-hosted assistants and research on mixture-of-experts.

Things to consider

Needs about 26 GB of memory at 4-bit despite its low active parameter count.
Superseded by newer models.
Download sizes refer to the default 4-bit quantized versions in the Ollama library; full-precision weights are larger.

Specifications are taken from the model card on Hugging Face, the Ollama library and the provider’s announcement. Benchmark results depend on evaluation settings; see how AI benchmarks work.

#Open Weight Models #Open Weights #Local AI #Mistral AI