MI

Magistral-Small-2506

MistralGGUF by unslothUpdated 2025-07-24
Download on Hugging Face
Parameters23.6B
Context40K
Layers40
Architecturellama
Smallest quant6.8 GB
Quick answer

Magistral-Small-2506 needs about 16.2 GB of memory at Q4_K_M (a 14.3 GB file) with an 8K context window, so it fits on a 24 GB GPU. This repository has 24 quants, from 6.8 GB (IQ2_XXS) to 47.2 GB (BF16). Longer context needs more: at 32K tokens the KV cache alone adds about 5.4 GB.

Magistral‑Small‑2506: 23.6B parameter reasoning model

Magistral‑Small‑2506 is a 23.6B parameter model built on Mistral Small 3.1 with added reasoning capabilities. It supports a 128k context window, though performance may degrade past 40k tokens. The model is licensed under Apache‑2.0 and can be run locally on a single RTX 4090 or a 32GB RAM MacBook when quantized.

Good at

  • Long‑chain reasoning
  • Multilingual support
  • Vision support in llama.cpp

Limits

  • Large context window may degrade past 40k
  • Requires quantization for local deployment
  • No explicit vision model size limits

Best for

  • Local reasoning tasks
  • Multilingual chat applications
  • Vision‑enabled inference with llama.cpp

Who it’s for. Developers needing a small, efficient reasoning model for local deployment. The model is suitable for applications that require long context handling and multilingual support while staying within the memory limits of a single high‑end GPU or a 32GB RAM machine.

Quants & memory

QuantFileEst. memoryvs VRAMResultDownload
IQ2_XXS6.8 GB 8.6 GB
Too big Get file →
IQ2_M8.2 GB 10.1 GB
Too big Get file →
Q2_K8.9 GB 10.7 GB
Too big Get file →
Q2_K_L9.0 GB 10.9 GB
Too big Get file →
Q2_K_XL9.3 GB 11.1 GB
Too big Get file →
IQ3_XXS9.4 GB 11.3 GB
Too big Get file →
Q3_K_S10.4 GB 12.2 GB
Too big Get file →
Q3_K_M11.5 GB 13.3 GB
Too big Get file →
Q3_K_XL11.9 GB 13.7 GB
Too big Get file →
IQ4_XS12.8 GB 14.6 GB
Too big Get file →
IQ4_NL13.5 GB 15.3 GB
Too big Get file →
Q4_013.5 GB 15.3 GB
Too big Get file →
Q4_K_S13.5 GB 15.4 GB
Too big Get file →
Q4_K_M14.3 GB 16.2 GB
Too big Get file →
Q4_K_XL14.5 GB 16.3 GB
Too big Get file →
Q4_114.9 GB 16.7 GB
Too big Get file →
Q5_K_S16.3 GB 18.1 GB
Too big Get file →
Q5_K_M16.8 GB 18.6 GB
Too big Get file →
Q5_K_XL16.8 GB 18.6 GB
Too big Get file →
Q6_K19.3 GB 21.2 GB
Too big Get file →
Q6_K_XL20.8 GB 22.6 GB
Too big Get file →
Q8_025.1 GB 26.9 GB
Too big Get file →
Q8_K_XL29.0 GB 30.8 GB
Too big Get file →
BF1647.2 GB 49.0 GB
Too big Get file →

For 8GB VRAM use IQ2_XXS (6.75GB) or IQ2_M (8.24GB). For 12GB VRAM IQ3_XXS (9.41GB) or Q3_K_S (10.4GB). For 24GB VRAM Q4_K_M (14.33GB) or Q5_K_S (16.3GB). Estimates are approximate.

What GPU do you need for Magistral-Small-2506?

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 IQ3_XXS (9.4 GB file, about 11.3 GB in use).
  • 16 GB VRAM: the best quant that fits is Q4_K_S (13.5 GB file, about 15.4 GB in use).
  • 24 GB VRAM: the best quant that fits is Q6_K_XL (20.8 GB file, about 22.6 GB in use).
  • 32 GB VRAM: the best quant that fits is Q8_K_XL (29.0 GB file, about 30.8 GB in use).
  • 48 GB VRAM: the best quant that fits is Q8_K_XL (29.0 GB file, about 30.8 GB in use).

Which Magistral-Small-2506 quant should you choose?

Q4_K_M is the usual starting point for Magistral-Small-2506: it balances file size and quality well. Here it is a 14.3 GB file.

The Q8_0 file is 25.1 GB, about 1.7 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 IQ2_XXS at 6.8 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
Parameters23.6B
Layers40
Trained context40,960 tokens
KV cache (fp16)160 KB per token

Source

Quantized byunsloth
FamilyMistral
Licenseapache-2.0
Downloads (HF)3,070
Last updated2025-07-24

Frequently asked questions about Magistral-Small-2506

How much VRAM does Magistral-Small-2506 need?

About 16.2 GB at Q4_K_M with an 8K context window. That is the 14.3 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 Magistral-Small-2506 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 Magistral-Small-2506 GGUF?

It depends on the quant: IQ2_XXS is 6.8 GB, IQ2_M is 8.2 GB, Q2_K is 8.9 GB, Q2_K_L is 9.0 GB, Q2_K_XL is 9.3 GB, IQ3_XXS is 9.4 GB and 18 more.

What context length does Magistral-Small-2506 support?

The GGUF header lists a trained context length of 40,960 tokens (40K). The KV cache grows with the context you use: about 1.3 GB at 8K and 5.4 GB at 32K.

Where can I download Magistral-Small-2506 GGUF?

The files are on Hugging Face in the unsloth/Magistral-Small-2506-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 Magistral-Small-2506 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 Magistral‑Small‑2506 on an RTX 4090?

Yes, the model fits on an RTX 4090 when quantized, for example with IQ4_K_M or Q5_K_S.

What is the maximum context length I should use?

Set the maximum model length to 40k tokens to avoid performance degradation.

Is the model licensed for commercial use?

Yes, it is released under the Apache‑2.0 license, allowing commercial and non‑commercial use.

Other GGUFs of Magistral-Small-2506

Similar models

Browse all →

More from unsloth

All by unsloth →

More Mistral models

All Mistral →

Other medium models

All medium models →