mistralai.Mistral Small 4 119B 2603 BF16 needs about 73.0 GB of memory at Q4_K_M (a 72.2 GB file) with an 8K context window, which is more than a 48 GB GPU holds. This repository has 6 quants, from 43.4 GB (Q2_K) to 126.5 GB (Q8_0). Longer context needs more: at 32K tokens the KV cache alone adds about 1.4 GB.
Mistral Small 4 119B – a multimodal, MoE LLM with 256k context
Mistral Small 4 119B is a hybrid model that combines instruction, reasoning, and devstral capabilities. It supports text and image input, function calls, and a 256k context window. The model is built on a deepseek2 architecture with 119B parameters and 128 experts, 4 active per token.
Good at
- Multimodal text & image tasks
- Large context reasoning
- Function calling
Limits
- Requires >40GB VRAM for Q2_K
- No vision support in GGUF quantized files
- License unspecified in metadata
Best for
- Chatbots needing long context
- Image captioning with text output
- Agentic workflows with JSON output
Who it’s for. Developers building advanced chat or agent systems that need large context and multimodal input. The model is suitable for research and production environments that can provide the necessary GPU memory.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| Q2_K | 43.4 GB (3 parts) | 44.2 GB | Too big | Get file → | |
| Q3_K_M | 56.9 GB (4 parts) | 57.8 GB | Too big | Get file → | |
| Q4_K_M | 72.2 GB (5 parts) | 73.0 GB | Too big | Get file → | |
| Q5_K_M | 84.5 GB (6 parts) | 85.4 GB | Too big | Get file → | |
| Q6_K | 97.7 GB (7 parts) | 98.5 GB | Too big | Get file → | |
| Q8_0 | 126.5 GB (9 parts) | 127.3 GB | Too big | Get file → |
For 8GB VRAM, the smallest quant is Q2_K at 43.4GB file size, but it still needs about 44.2GB VRAM at 8k context – not feasible. Q3_K_M requires 56.94GB file size and 57.8GB VRAM. The only practical option for limited memory is to use a higher precision quant like Q4_K_M or Q5_K_M if you can allocate 73GB or 85.4GB VRAM respectively. Memory estimates are approximate.
What GPU do you need for mistralai.Mistral Small 4 119B 2603 BF16?
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: no quant fits at an 8K context.
- 16 GB VRAM: no quant fits at an 8K context.
- 24 GB VRAM: no quant fits at an 8K context.
- 32 GB VRAM: no quant fits at an 8K context.
- 48 GB VRAM: the best quant that fits is Q2_K (43.4 GB file, about 44.2 GB in use).
Which mistralai.Mistral Small 4 119B 2603 BF16 quant should you choose?
Q4_K_M is the usual starting point for mistralai.Mistral Small 4 119B 2603 BF16: it balances file size and quality well. Here it is a 72.2 GB file.
The Q8_0 file is 126.5 GB, about 1.8 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 43.4 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
Source
Frequently asked questions about mistralai.Mistral Small 4 119B 2603 BF16
How much VRAM does mistralai.Mistral Small 4 119B 2603 BF16 need?
About 73.0 GB at Q4_K_M with an 8K context window. That is the 72.2 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 mistralai.Mistral Small 4 119B 2603 BF16 on an 8 GB GPU?
Not at an 8K context. The smallest file needs about 44.2 GB, so you would need a larger GPU or CPU offloading.
What is the file size of mistralai.Mistral Small 4 119B 2603 BF16 GGUF?
It depends on the quant: Q2_K is 43.4 GB, Q3_K_M is 56.9 GB, Q4_K_M is 72.2 GB, Q5_K_M is 84.5 GB, Q6_K is 97.7 GB, Q8_0 is 126.5 GB.
What context length does mistralai.Mistral Small 4 119B 2603 BF16 support?
The GGUF header lists a trained context length of 1,048,576 tokens (1024K). The KV cache grows with the context you use: about 0.3 GB at 8K and 1.4 GB at 32K.
Where can I download mistralai.Mistral Small 4 119B 2603 BF16 GGUF?
The files are on Hugging Face in the DevQuasar/mistralai.Mistral-Small-4-119B-2603-GGUF repository. The download buttons on this page open the file on Hugging Face. ggufmodels.com does not host any model files.
Can I run Mistral Small 4 on 8GB VRAM?
No. Even the smallest quant Q2_K requires about 44.2GB VRAM at 8k context, which exceeds 8GB.
Does the model support vision?
Yes, it accepts image input and provides text output, but the GGUF quantized files do not include vision weights.
What is the context length?
The model supports a 256k context window, trained on 1,048,576 tokens.