Cyberax AI Playbook
cyberax.com
Model · Mistral AI · Text

Mixtral 8x7B

Mistral's mixture-of-experts model. 8 experts × 7B params with 2 active per token — punches above its compute weight for inference speed.

Modality
Text
License
Apache 2.0 (Open)
Parameter size
8x7B (47B total)
Context window
32,768 tokens
Released
December 11, 2023
Last verified
May 10, 2026
Runs locally
Yes

Strengths

  • Apache 2.0 — fully open-licensed
  • MoE architecture — only 2 of 8 experts active per token
  • Strong multilingual coverage

Weaknesses

  • Needs ~90GB VRAM at full precision (or quantization)
  • Older architecture — newer dense models often match its quality

Try it

WhereTypeNotes
Hugging Face weights Apache 2.0
OpenRouter hosted-api API key required
Ollama local ollama run mixtral

Change log

  • — Initial entry.
  • — Marked deprecated after verification that Mixtral 8x7B no longer appears in Mistral's current model overview.