Mistral AI · 47B (12.9B active) · 專家混合（MoE）
MoE with 12.9B active params 看看你的 GPU 或 Mac 跑不跑得動 Mixtral 8x7B——最低 26.3 GB，建議 43.8 GB。
2023-1232K context
專家混合（MoE）
專家總数： 8
啟用專家： 2
啟用参数： 12.9B
| 量化 | 位元 | VRAM | 品質 | 状態 |
|---|---|---|---|---|
| Q2_K | 2 | 15.5 GB | low | — |
| Q3_K_M | 3 | 21.6 GB | moderate | — |
| Q4_K_M | 4 | 24.6 GB | good | — |
| Q5_K_M | 5 | 30.6 GB | good | — |
| Q6_K | 6 | 36.6 GB | excellent | — |
| Q8_0 | 8 | 48.6 GB | excellent | — |
| F16 | 16 | 96.8 GB | lossless | — |
關於這個模型
The Mixtral large Language Models (LLM) are a set of pretrained generative Sparse Mixture of Experts.
Sizes
mixtral:8x22bmixtral:8x7b
Mixtral 8x22b
ollama run mixtral:8x22b
Mixtral 8x22B sets a new standard for performance and efficiency within the AI community. It is a sparse Mixture-of-Experts (SMoE) model that uses only 39B active parameters out of 141B, offering unparalleled cost efficiency for its size.
Mixtral 8x22B comes with the following strengths:
- It is fluent in English, French, Italian, German, and Spanish
- It has strong maths and coding capabilities
- It is natively capable of function calling
- 64K tokens context window allows precise information recall from large documents
References
Can I run Mixtral 8x7B locally?
- Can I run Mixtral 8x7B locally?
- Mixtral 8x7B needs about 26.3 GB of memory at a minimum and 43.8 GB recommended. Open this page to grade it against your GPU or Mac, then run it with runai, Ollama or LM Studio.
- How much VRAM does Mixtral 8x7B need?
- At Q4_K_M, Mixtral 8x7B uses about 24.6 GB of VRAM. Higher quants need more memory; lower quants fit tighter cards with a quality tradeoff.