How much memory do I need to run OmniVoice (Q4_K_M)?

The file size is 0.25 GB (measured). With a 8,192-token context, the estimated memory including the KV cache is 2.04 GB. The formula is published at https://localllm.az-arte.com/en/guide/memory-calculation.

Does OmniVoice run in 8 GB, 12 GB, 16 GB or 24 GB of VRAM?

8 GB: up to F32 (est. 4.24 GB); 12 GB: up to F32 (est. 4.24 GB); 16 GB: up to F32 (est. 4.24 GB); 24 GB: up to F32 (est. 4.24 GB) (estimated at 8,192 tokens with an f16 KV cache, files from the most-downloaded source).

k2-fsa

OmniVoice

Speech / audio

Parameters: 0.6B
Max context length: 40,960
Architecture: omnivoice-lm
License: Non-commercial

Base model: k2-fsa/OmniVoice
License: cc-by-nc-4.0

Memory needed to run Q4_K_M at 8K tokens

2.04 GB Estimated

File size 0.25 GB Measured

VRAM it fits in (rough guide)

8 GB fits
12 GB fits
16 GB fits
24 GB fits
32 GB fits
48 GB fits

See devices and models for 4 GB

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

Data as of Nov 11, 2026

About OmniVoice

OmniVoice is a GGUF model based on k2-fsa/OmniVoice on Hugging Face. GGUF files are published by Serveurperso/OmniVoice-GGUF. It has about 613M parameters, the omnivoice-lm architecture, a maximum context length of 40,960 tokens (values written in the GGUF).

There are 4 quantizations, with file sizes from 0.25 GB (Q4_K_M) to 2.46 GB (F32) (actual sizes on Hugging Face).

Running Q4_K_M (0.25 GB) with an 8,192-token context needs an estimated 2.04 GB of memory, and an estimated 4.86 GB at 32,768 tokens. For 8,192 tokens, a GPU with 4 GB or more of VRAM (such as Arc A310, Arc A370M), or a Mac with 4 GB or more of unified memory, is a rough guide (macOS caps how much memory the GPU can use, so leave some headroom).

According to Serveurperso/OmniVoice-GGUF, the license is Creative Commons Attribution-NonCommercial 4.0 (CC BY-NC 4.0), which does not allow commercial use (limited to non-commercial use such as research or personal use). 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

Serveurperso/OmniVoice-GGUF

Converted and published by a third party About this label

Last checked: Nov 11, 2026

Architecture: omnivoice-lm; Not supported by mainline llama.cpp (as of b12025)
Parameters: 0.6B
Max context length: 40,960
License: cc-by-nc-4.0 Non-commercial

File size and memory per quantization for OmniVoice (Serveurperso/OmniVoice-GGUF)
Quantization	File size Measured	Quality guide	4,096	8,192	16,384	32,768	40,960 max
Q4_K_M omnivoice-tokenizer-Q4_K_M	0.25 GB	Small	1.57 GB	2.04 GB	2.98 GB	4.86 GB	5.80 GB
Q8_0 omnivoice-tokenizer-Q8_0	0.29 GB	Nearly identical	1.61 GB	2.08 GB	3.02 GB	4.89 GB	5.83 GB
BF16 omnivoice-tokenizer-BF16	0.37 GB	Original precision	1.69 GB	2.16 GB	3.10 GB	4.98 GB	5.92 GB
Q4_K_M omnivoice-base-Q4_K_M	0.41 GB	Small	1.72 GB	2.19 GB	3.13 GB	5.01 GB	5.95 GB
Q8_0 omnivoice-base-Q8_0	0.66 GB	Nearly identical	1.97 GB	2.44 GB	3.38 GB	5.26 GB	6.20 GB
F32 omnivoice-tokenizer-F32	0.73 GB	Original precision	2.05 GB	2.52 GB	3.46 GB	5.34 GB	6.28 GB
BF16 omnivoice-base-BF16	1.23 GB	Original precision	2.55 GB	3.02 GB	3.96 GB	5.84 GB	6.78 GB
F32 omnivoice-base-F32	2.46 GB	Original precision	3.77 GB	4.24 GB	5.18 GB	7.06 GB	8.00 GB

This GGUF format (omnivoice-lm) 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 (0.25 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
Arc A370M 4 GB	112 GB/s	266–355	444
Arc A770 (8GB) 8 GB	512 GB/s	1,217–1,622	2,028
GeForce RTX 5090 32 GB	1,792 GB/s	4,259–5,678	7,098
Apple M2 8 GB	100 GB/s	238–317	396
Apple M2 Pro 16 GB	200 GB/s	475–634	792
Apple M5 Ultra 96 GB	1,200 GB/s	2,852–3,802	4,753

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

Creative Commons Attribution-NonCommercial 4.0 (CC BY-NC 4.0) Non-commercial

You may use and modify it with attribution for non-commercial purposes only. Commercial use is not allowed.

The original model's license has not been fetched yet, so this is based on the GGUF repository's label. 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 OmniVoice (Q4_K_M)?: The file size is 0.25 GB (measured). With a 8,192-token context, the estimated memory including the KV cache is 2.04 GB. The formula is published at https://localllm.az-arte.com/en/guide/memory-calculation.
Can OmniVoice be used commercially?: The license is Creative Commons Attribution-NonCommercial 4.0 (CC BY-NC 4.0) (No commercial use).You may use and modify it with attribution for non-commercial purposes only. Commercial use is not allowed. Check the original license text before deciding.
Does OmniVoice run in 8 GB, 12 GB, 16 GB or 24 GB of VRAM?: 8 GB: up to F32 (est. 4.24 GB); 12 GB: up to F32 (est. 4.24 GB); 16 GB: up to F32 (est. 4.24 GB); 24 GB: up to F32 (est. 4.24 GB) (estimated at 8,192 tokens with an f16 KV cache, files from the most-downloaded source).