Mistral Magistral Devstral Instruct FUSED CODER Reasoning 36B needs about 24.2 GB of memory at Q4_K_M (a 21.7 GB file) with an 8K context window, so it fits on a 32 GB GPU. This repository has 24 quants, from 0.0 GB to 29.4 GB (Q6_K). Longer context needs more: at 32K tokens the KV cache alone adds about 8.3 GB.
Mistral 36B model for coding and reasoning
Mistral Magistral Devstral Instruct FUSED CODER Reasoning 36B is a 35.8B‑parameter llama architecture model. It was trained on 131072 context tokens and released under the apache-2.0 license. The model supports coding and reasoning tasks.
Good at
- Coding
- Reasoning
- Large context handling
Limits
- No vision support
- Requires large VRAM for full precision
- License is apache-2.0
Best for
- Developers building code assistants
- Researchers testing reasoning models
- Educational tools for programming
Who it’s for. The model is suitable for developers and researchers who need a large language model focused on coding and reasoning. It is also useful for educational applications that require detailed code generation and logical analysis.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| — | 0.0 GB | 2.6 GB | Fits | Get file → | |
| IQ1_S | 7.8 GB | 10.4 GB | Too big | Get file → | |
| IQ1_M | 8.5 GB | 11.1 GB | Too big | Get file → | |
| IQ2_XXS | 9.8 GB | 12.3 GB | Too big | Get file → | |
| IQ2_XS | 10.8 GB | 13.4 GB | Too big | Get file → | |
| IQ2_S | 11.2 GB | 13.7 GB | Too big | Get file → | |
| IQ2_M | 12.1 GB | 14.7 GB | Too big | Get file → | |
| Q2_K_S | 12.4 GB | 15.0 GB | Too big | Get file → | |
| Q2_K | 13.4 GB | 15.9 GB | Too big | Get file → | |
| IQ3_XXS | 14.0 GB | 16.5 GB | Too big | Get file → | |
| IQ3_XS | 14.9 GB | 17.5 GB | Too big | Get file → | |
| Q3_K_S | 15.7 GB | 18.2 GB | Too big | Get file → | |
| IQ3_S | 15.7 GB | 18.3 GB | Too big | Get file → | |
| IQ3_M | 16.0 GB | 18.6 GB | Too big | Get file → | |
| Q3_K_M | 17.3 GB | 19.9 GB | Too big | Get file → | |
| Q3_K_L | 18.8 GB | 21.3 GB | Too big | Get file → | |
| IQ4_XS | 19.3 GB | 21.9 GB | Too big | Get file → | |
| Q4_0 | 20.4 GB | 23.0 GB | Too big | Get file → | |
| Q4_K_S | 20.5 GB | 23.0 GB | Too big | Get file → | |
| Q4_K_M | 21.7 GB | 24.2 GB | Too big | Get file → | |
| Q4_1 | 22.5 GB | 25.1 GB | Too big | Get file → | |
| Q5_K_S | 24.7 GB | 27.3 GB | Too big | Get file → | |
| Q5_K_M | 25.4 GB | 28.0 GB | Too big | Get file → | |
| Q6_K | 29.4 GB | 32.0 GB | Too big | Get file → |
For 8 GB VRAM use IQ1_S (10.4 GB memory) or Q2_K_S (15 GB memory). For 16 GB VRAM use IQ2_S (13.7 GB memory) or Q3_K_S (18.2 GB memory). For 32 GB VRAM use Q4_K_M (24.2 GB memory) or Q5_K_M (28 GB memory). Estimates are approximate.
What GPU do you need for Mistral Magistral Devstral Instruct FUSED CODER Reasoning 36B?
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: the best quant that fits is the main file (0.0 GB file, about 2.6 GB in use).
- 6 GB VRAM: the best quant that fits is the main file (0.0 GB file, about 2.6 GB in use).
- 8 GB VRAM: the best quant that fits is the main file (0.0 GB file, about 2.6 GB in use).
- 12 GB VRAM: the best quant that fits is IQ1_M (8.5 GB file, about 11.1 GB in use).
- 16 GB VRAM: the best quant that fits is Q2_K (13.4 GB file, about 15.9 GB in use).
- 24 GB VRAM: the best quant that fits is Q4_K_S (20.5 GB file, about 23.0 GB in use).
- 32 GB VRAM: the best quant that fits is Q6_K (29.4 GB file, about 32.0 GB in use).
Which Mistral Magistral Devstral Instruct FUSED CODER Reasoning 36B quant should you choose?
Q4_K_M is the usual starting point for Mistral Magistral Devstral Instruct FUSED CODER Reasoning 36B: it balances file size and quality well. Here it is a 21.7 GB file.
If memory is tight, the smallest file is the first one in the table at 0.0 GB, with the quality trade-offs that come with fewer bits.
New to quant names? Read GGUF quantization explained.
Specifications
Architecture from the GGUF header
Source
Frequently asked questions about Mistral Magistral Devstral Instruct FUSED CODER Reasoning 36B
How much VRAM does Mistral Magistral Devstral Instruct FUSED CODER Reasoning 36B need?
About 24.2 GB at Q4_K_M with an 8K context window. That is the 21.7 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 Mistral Magistral Devstral Instruct FUSED CODER Reasoning 36B on an 8 GB GPU?
Yes, with the smaller files: the 0.0 GB file needs about 2.6 GB at 8K context. Larger quants in this repository do not fit in 8 GB.
What is the file size of Mistral Magistral Devstral Instruct FUSED CODER Reasoning 36B GGUF?
It depends on the quant: file is 0.0 GB, IQ1_S is 7.8 GB, IQ1_M is 8.5 GB, IQ2_XXS is 9.8 GB, IQ2_XS is 10.8 GB, IQ2_S is 11.2 GB and 18 more.
What context length does Mistral Magistral Devstral Instruct FUSED CODER Reasoning 36B support?
The GGUF header lists a trained context length of 131,072 tokens (128K). The KV cache grows with the context you use: about 2.1 GB at 8K and 8.3 GB at 32K.
Where can I download Mistral Magistral Devstral Instruct FUSED CODER Reasoning 36B GGUF?
The files are on Hugging Face in the mradermacher/Mistral-Magistral-Devstral-Instruct-FUSED-CODER-Reasoning-36B-i1-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 Mistral Magistral Devstral Instruct FUSED CODER Reasoning 36B use?
Hugging Face lists the license as apache-2.0. 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 this model on 8 GB VRAM?
Use IQ1_S or Q2_K_S quantizations; they require about 10.4 GB or 15 GB of VRAM at 8k context.
Does the model support image input?
No, the model has no vision capabilities.
What license applies to this model?
The model is released under the apache-2.0 license.