How much memory do I need to run Voxtral-Small-24B-2507 (Q4_K_M)?

The file size is 14.3 GB (measured). With a 8,192-token context, the estimated memory including the KV cache is 16.4 GB. The formula is published at https://localllm.az-arte.com/en/guide/memory-calculation.

Does Voxtral-Small-24B-2507 run in 8 GB, 12 GB, 16 GB or 24 GB of VRAM?

8 GB: no quantization fits; 12 GB: no quantization fits; 16 GB: up to Q4_K_M (est. 16.4 GB); 24 GB: up to Q6_K (est. 22.1 GB) (estimated at 8,192 tokens with an f16 KV cache, files from the most-downloaded source).

mistralai

Voxtral-Small-24B-2507

Speech / audio

Parameters: 24.3B
Architecture: voxtral
License: Commercial OK

Base model: mistralai/Voxtral-Small-24B-2507
License: apache-2.0

Memory needed to run Q4_K_M at 8K tokens

16.4 GB Estimated

File size 14.3 GB Measured

VRAM it fits in (rough guide)

8 GB does not fit
12 GB does not fit
16 GB fits
24 GB fits
32 GB fits
48 GB fits

See devices and models for 16 GB

Macs cap how much memory the GPU can use, so leave headroom in unified memory.

Data as of Nov 11, 2026

About Voxtral-Small-24B-2507

Voxtral-Small-24B-2507 is a GGUF model based on mistralai/Voxtral-Small-24B-2507 on Hugging Face. GGUF files are published by handy-computer/Voxtral-Small-24B-2507-gguf. It has about 24.3B parameters, the voxtral architecture (values written in the GGUF).

There are 6 quantizations, with file sizes from 14.3 GB (Q4_K_M) to 48.5 GB (F16) (actual sizes on Hugging Face).

Running Q4_K_M (14.3 GB) with an 8,192-token context needs an estimated 16.4 GB of memory, and an estimated 20.5 GB at 32,768 tokens. For 8,192 tokens, a GPU with 16 GB or more of VRAM (such as GeForce RTX 5080, GeForce RTX 5070 Ti), or a Mac with 16 GB or more of unified memory, is a rough guide (macOS caps how much memory the GPU can use, so leave some headroom).

The original model is licensed under Apache License 2.0, which allows commercial use (keep the license text and copyright notice when redistributing). Check the original license for the terms of use.

File size and estimated memory per quantization

Up to 8 GB
Up to 16 GB
Up to 24 GB
Up to 48 GB
Up to 128 GB
More

File sizes are the actual values on Hugging Face. Memory is estimated for each context length (KV cache f16). The formula is in How memory is calculated.

All memory figures are Estimated (KV cache f16). It should run on a setup with at least this much VRAM (or Mac unified memory). Click a value to see the devices and models that fit that amount (Find by GPU / Mac).

About the "quality guide" (how far a quantization drifts from the original)

For each quantization type, how far the output drifts from the original model, grouped into six levels from values llama.cpp measured on Llama-3-8B.

The value is mean KLD (the difference between the original and quantized model in next-token probabilities); smaller means closer to the original. Quantizations made with an imatrix (importance data) come out slightly smaller. Levels use the value without an imatrix (or with one, for types that only have that).

These values were not measured on this model. Larger models and MoE models tend to drift less with the same type. Distributor-specific mixed types (such as UD-Q4_K_XL) cannot be matched by type name alone, so they show "-".

Quantization	Quality guide	KLD (no imatrix)	KLD (imatrix)	Size (Llama-3-8B)
Q8_0	Nearly identical	0.0014	-	7.96 GiB
Q6_K	Very small	0.0055	-	6.14 GiB
Q5_K_M	Very small	0.0108	-	5.33 GiB
Q5_K_S	Small	0.0166	-	5.21 GiB
Q5_1	Small	0.0180	-	5.65 GiB
Q5_0	Small	0.0222	-	5.21 GiB
Q4_K_M	Small	0.0313	0.0282	4.58 GiB
Q4_K_S	Medium	0.0431	0.0320	4.37 GiB
IQ4_NL	Small	-	0.0357	4.35 GiB
IQ4_XS	Small	-	0.0363	4.14 GiB
Q4_1	Medium	0.0717	-	4.78 GiB
Q4_0	Medium	0.0719	-	4.34 GiB
Q3_K_L	Medium	0.0872	0.0731	4.03 GiB
Q3_K_M	Medium	0.1019	0.0844	3.74 GiB
IQ3_M	Medium	-	0.1025	3.53 GiB
IQ3_S	Medium	-	0.1113	3.42 GiB
IQ3_XS	Large	-	0.1387	3.28 GiB
IQ3_XXS	Large	-	0.1836	3.05 GiB
Q3_K_S	Large	0.2282	0.1998	3.41 GiB
IQ2_M	Large	-	0.3260	2.74 GiB
Q2_K	Very large	0.4451	0.3322	2.96 GiB
Q2_K_S	Very large	-	0.4034	2.96 GiB
IQ2_S	Very large	-	0.4392	2.56 GiB
IQ2_XS	Very large	-	0.5463	2.43 GiB
IQ2_XXS	Very large	-	0.8120	2.24 GiB
IQ1_M	Very large	-	1.3931	2.01 GiB
IQ1_S	Very large	-	2.2546	1.88 GiB

Source: llama.cpp (MIT), measured on Llama-3-8B against FP16, Wikitext-2, revision f364eb6f. Original table · Reading and choosing quantizations

GGUF source

handy-computer/Voxtral-Small-24B-2507-gguf

Converted and published by a third party About this label

Last checked: Nov 11, 2026

Architecture: voxtral; Not supported by mainline llama.cpp (as of b12025)
Parameters: 24.3B
License: apache-2.0 Commercial OK

File size and memory per quantization for Voxtral-Small-24B-2507 (handy-computer/Voxtral-Small-24B-2507-gguf)
Quantization	File size Measured	Quality guide	4,096	8,192	16,384	32,768	65,536	131,072
Q4_K_M	14.3 GB	Small	15.8 GB	16.4 GB	17.8 GB	20.5 GB	25.8 GB	36.6 GB
Q5_K_M	17.1 GB	Very small	18.6 GB	19.3 GB	20.6 GB	23.3 GB	28.7 GB	39.4 GB
Q6_K	19.9 GB	Very small	21.4 GB	22.1 GB	23.4 GB	26.1 GB	31.5 GB	42.2 GB
Q8_0	25.8 GB	Nearly identical	27.3 GB	28.0 GB	29.3 GB	32.0 GB	37.4 GB	48.1 GB
BF16	48.5 GB	Original precision	50.0 GB	50.7 GB	52.0 GB	54.7 GB	60.1 GB	70.8 GB
F16	48.5 GB	Original precision	50.0 GB	50.7 GB	52.0 GB	54.7 GB	60.1 GB	70.8 GB

This GGUF format (voxtral) does not exist in mainline llama.cpp (as of b12025). Ollama, LM Studio and mainline llama.cpp cannot load it; use the dedicated tool described by the source.

Generation speed on common GPUs and Macs (estimated)

A rough guide to how fast Q4_K_M (14.3 GB) generates text. Each token reads the model weights once, so the upper limit is memory bandwidth ÷ file size. The guide is 60–80% of that, where most published measurements fall.

Device	Memory bandwidth	Guide (tokens/s) Estimated	Upper limit
Radeon RX 7600 XT 16 GB	288 GB/s	12–16	20
Radeon RX 9070 16 GB	640 GB/s	27–36	45
GeForce RTX 5090 32 GB	1,792 GB/s	75–100	125
Apple M2 24 GB	100 GB/s	4.2–5.6	7.0
Apple M2 Pro 32 GB	200 GB/s	8.4–11	14
Apple M5 Ultra 96 GB	1,200 GB/s	50–67	84

Estimates only. Prompt processing speed (limited by compute) is not included. Longer contexts and offloading part of the model to the CPU make it slower. Macs are shown only when installed memory is at least 4/3 of what is needed. Devices without published bandwidth (some GeForce cards, etc.) are not shown.

License and commercial use

Apache License 2.0 Commercial OK

Allows commercial use, modification and redistribution. When redistributing, keep the license text and copyright notice, and note changes in modified files.

Based on the license on the original model's Hugging Face page. Original license text (checked Oct 18, 2026)

The summary states facts this site read from the original text and is not legal advice. Always check the original license and its latest terms before use. All licenses

Frequently asked questions

How much memory do I need to run Voxtral-Small-24B-2507 (Q4_K_M)?: The file size is 14.3 GB (measured). With a 8,192-token context, the estimated memory including the KV cache is 16.4 GB. The formula is published at https://localllm.az-arte.com/en/guide/memory-calculation.
Can Voxtral-Small-24B-2507 be used commercially?: The original model's license is Apache License 2.0 (Commercial use allowed).Allows commercial use, modification and redistribution. When redistributing, keep the license text and copyright notice, and note changes in modified files. Check the original license text before deciding.
Does Voxtral-Small-24B-2507 run in 8 GB, 12 GB, 16 GB or 24 GB of VRAM?: 8 GB: no quantization fits; 12 GB: no quantization fits; 16 GB: up to Q4_K_M (est. 16.4 GB); 24 GB: up to Q6_K (est. 22.1 GB) (estimated at 8,192 tokens with an f16 KV cache, files from the most-downloaded source).