MI

Mixtral 11Bx2 MoE 19B

MistralGGUF by TheBlokeUpdated 2024-01-05
Download on Hugging Face
Parameters19.2B
Context4K
Layers48
Architecturellama
Smallest quant6.5 GB
Quick answer

Mixtral 11Bx2 MoE 19B needs about 13.0 GB of memory at Q4_K_M (a 10.9 GB file) with an 8K context window, so it fits on a 16 GB GPU. This repository has 12 quants, from 6.5 GB (Q2_K) to 20.4 GB (Q8_0).

Mixtral 11Bx2 MoE 19B – 19.2B parameter Mixture of Experts model

Mixtral 11Bx2 MoE 19B is a 19.2B parameter model built on the Mistral family. It uses a mixture of experts architecture and supports a 4096 token context window. The model is released under a cc‑by‑nc‑4.0 license.

Good at

  • Large context handling
  • Mixture of experts inference
  • High parameter count

Limits

  • No vision support
  • Requires GPU for efficient use
  • cc‑by‑nc‑4.0 license limits commercial use

Best for

  • Research prototyping
  • Complex text generation
  • Fine‑tuning on specialized data

Who it’s for. Engineers and researchers who need a large, expert‑based language model for non‑commercial projects. The model is suitable for local deployment with GPU acceleration.

Quants & memory

QuantFileEst. memoryvs VRAMResultDownload
Q2_K6.5 GB 7.8 GB
Tight Get file →
Q3_K_S8.3 GB 9.6 GB
Too big Get file →
Q3_K_M8.4 GB 9.7 GB
Too big Get file →
Q3_K_L8.6 GB 9.9 GB
Too big Get file →
Q4_010.8 GB 12.1 GB
Too big Get file →
Q4_K_S10.8 GB 12.1 GB
Too big Get file →
Q4_K_M10.9 GB 12.2 GB
Too big Get file →
Q5_013.2 GB 14.5 GB
Too big Get file →
Q5_K_S13.2 GB 14.5 GB
Too big Get file →
Q5_K_M13.2 GB 14.5 GB
Too big Get file →
Q6_K15.7 GB 17.0 GB
Too big Get file →
Q8_020.4 GB 21.7 GB
Too big Get file →

For 8 GB VRAM, use Q4_K_M (13 GB file, ~13 GB memory). For 12 GB VRAM, Q4_0 (10.83 GB file, ~12.9 GB memory). For 16 GB VRAM, Q5_0 (13.21 GB file, ~15.3 GB memory). Estimates are approximate.

What GPU do you need for Mixtral 11Bx2 MoE 19B?

These figures add the file size, an fp16 KV cache at an 8K context window (worked out from this model’s own header) and about 0.5 GB for the runtime. They are estimates, so leave a little headroom, and expect to use less context or a smaller quant if you are right at the limit.

  • 4 GB VRAM: no quant fits at an 8K context.
  • 6 GB VRAM: no quant fits at an 8K context.
  • 8 GB VRAM: no quant fits at an 8K context.
  • 12 GB VRAM: the best quant that fits is Q3_K_L (8.6 GB file, about 10.7 GB in use).
  • 16 GB VRAM: the best quant that fits is Q5_K_M (13.2 GB file, about 15.3 GB in use).
  • 24 GB VRAM: the best quant that fits is Q8_0 (20.4 GB file, about 22.5 GB in use).

Which Mixtral 11Bx2 MoE 19B quant should you choose?

Q4_K_M is the usual starting point for Mixtral 11Bx2 MoE 19B: it balances file size and quality well. Here it is a 10.9 GB file.

The Q8_0 file is 20.4 GB, about 1.9 times the size of Q4_K_M. It keeps the weights closer to the original, which only matters if you have the memory to spare.

If memory is tight, the smallest file is Q2_K at 6.5 GB, with the quality trade-offs that come with fewer bits. Use it only when memory is very tight. Expect a clear drop in quality, which is usually worse on small models and easier to live with on large ones.

New to quant names? Read GGUF quantization explained.

Specifications

Architecture from the GGUF header

Architecturellama
Parameters19.2B (approximate, mixture of experts)
Layers48
Trained context4,096 tokens
KV cache (fp16)192 KB per token

Source

Quantized byTheBloke
FamilyMistral
Downloads (HF)2,199
Last updated2024-01-05

Frequently asked questions about Mixtral 11Bx2 MoE 19B

How much VRAM does Mixtral 11Bx2 MoE 19B need?

About 13.0 GB at Q4_K_M with an 8K context window. That is the 10.9 GB file, plus the KV cache and roughly 0.5 GB of runtime overhead. Smaller quants need less and longer context needs more.

Can I run Mixtral 11Bx2 MoE 19B on an 8 GB GPU?

Not at an 8K context. The smallest file needs about 8.6 GB, so you would need a larger GPU or CPU offloading.

What is the file size of Mixtral 11Bx2 MoE 19B GGUF?

It depends on the quant: Q2_K is 6.5 GB, Q3_K_S is 8.3 GB, Q3_K_M is 8.4 GB, Q3_K_L is 8.6 GB, Q4_0 is 10.8 GB, Q4_K_S is 10.8 GB and 6 more.

What context length does Mixtral 11Bx2 MoE 19B support?

The GGUF header lists a trained context length of 4,096 tokens (4K). The KV cache grows with the context you use: about 1.6 GB at 8K and 0.8 GB at 4K.

Where can I download Mixtral 11Bx2 MoE 19B GGUF?

The files are on Hugging Face in the TheBloke/Mixtral_11Bx2_MoE_19B-GGUF repository. The download buttons on this page open the file on Hugging Face. ggufmodels.com does not host any model files.

What license does Mixtral 11Bx2 MoE 19B use?

Hugging Face lists the license as cc-by-nc-4.0. Licenses can differ between the quantized files and the original model, so check the repository and the base model before commercial use.

Can I run Mixtral 11Bx2 MoE 19B on 8 GB VRAM?

Use the Q4_K_M quantised file; it requires about 13 GB of VRAM at 8 k context.

Does this model support image inputs?

No, the model has no vision capabilities.

Is the model free for commercial use?

No, it is licensed under cc‑by‑nc‑4.0, which prohibits commercial use.

Other GGUFs of Mixtral_11Bx2_MoE_19B

Similar models

Browse all →

More from TheBloke

All by TheBloke →

More Mistral models

All Mistral →

Other medium models

All medium models →