How much memory do I need to run POCKET-26B (Q4_K_M)?

The file size is 16.8 GB (measured). With a 8,192-token context, the estimated memory including the KV cache is 18.4 GB. The formula is published at https://localllm.az-arte.com/en/guide/memory-calculation.

Does POCKET-26B run in 8 GB, 12 GB, 16 GB or 24 GB of VRAM?

8 GB: no quantization fits; 12 GB: up to Q2_K (est. 12.7 GB); 16 GB: up to Q2_K (est. 12.7 GB); 24 GB: up to Q4_K_M (est. 18.4 GB) (estimated at 8,192 tokens with an f16 KV cache, files from the most-downloaded source).

What is the command to run POCKET-26B with Ollama?

Run "ollama run hf.co/FINAL-Bench/POCKET-26B-GGUF:Q4_K_M" to use the GGUF on Hugging Face directly. To use another quantization, replace the name at the end.

FINAL-Bench

POCKET-26B

Parameters: 25.2B
Max context length: 262,144
Architecture: gemma4
License: Commercial OK

Base model: google/gemma-4-26B-A4B-it
License: apache-2.0
llama.cpp: b8637+

Memory needed to run Q4_K_M at 8K tokens

18.4 GB Estimated

File size 16.8 GB Measured

VRAM it fits in (rough guide)

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

See devices and models for 24 GB

The smallest, Q2_K, needs an estimated 12.7 GB at 8K.

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

Data as of Nov 11, 2026

About POCKET-26B

POCKET-26B is a GGUF model based on google/gemma-4-26B-A4B-it on Hugging Face. GGUF files are published by FINAL-Bench/POCKET-26B-GGUF. It has about 25.2B parameters, the gemma4 architecture, a maximum context length of 262,144 tokens (values written in the GGUF).

There are 2 quantizations, with file sizes from 11.1 GB (Q2_K) to 16.8 GB (Q4_K_M) (actual sizes on Hugging Face).

Running Q4_K_M (16.8 GB) with an 8,192-token context needs an estimated 18.4 GB of memory, and an estimated 18.9 GB at 32,768 tokens. For 8,192 tokens, a GPU with 24 GB or more of VRAM (such as GeForce RTX 3090 Ti, GeForce RTX 4090), or a Mac with 24 GB or more of unified memory, is a rough guide (macOS caps how much memory the GPU can use, so leave some headroom). The smallest, Q2_K (11.1 GB), needs an estimated 12.7 GB at 8,192 tokens.

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

FINAL-Bench/POCKET-26B-GGUF

Published by the original developer About this label

Last checked: Nov 11, 2026

Architecture: gemma4; llama.cpp b8637 and later
Parameters: 25.2B
Max context length: 262,144
License: apache-2.0 Commercial OK

File size and memory per quantization for POCKET-26B (FINAL-Bench/POCKET-26B-GGUF)
Quantization	File size Measured	Quality guide	4,096	8,192	16,384	32,768	65,536	131,072	262,144 max
Q2_K	11.1 GB	Very large	12.6 GB	12.7 GB	12.8 GB	13.2 GB	13.9 GB	15.2 GB	17.9 GB
Q4_K_M	16.8 GB	Small	18.3 GB	18.4 GB	18.5 GB	18.9 GB	19.5 GB	20.9 GB	23.6 GB

Ollama  ollama run hf.co/FINAL-Bench/POCKET-26B-GGUF:Q4_K_M

llama.cpp  llama-server -hf FINAL-Bench/POCKET-26B-GGUF:Q4_K_M

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

How to run

Mainline llama.cpp can load this model's format (gemma4) from b8637 (Apr 2, 2026) onward. Ollama and LM Studio bundle llama.cpp internally, so an older app fails to load it with errors such as "unknown model architecture". If that happens, update the app.

Example commands using Q4_K_M. To use another quantization, replace the name with one from the table above. How to run local LLMs

Ollama

Run it by pointing directly at the GGUF on Hugging Face.

ollama run hf.co/FINAL-Bench/POCKET-26B-GGUF:Q4_K_M

llama.cpp

Pass the repository and quantization name to llama-server (or llama-cli) with -hf.

llama-server -hf FINAL-Bench/POCKET-26B-GGUF:Q4_K_M

LM Studio

Paste FINAL-Bench/POCKET-26B-GGUF into the search box, download it, and choose the quantization from the file list.

Frequently asked questions

How much memory do I need to run POCKET-26B (Q4_K_M)?: The file size is 16.8 GB (measured). With a 8,192-token context, the estimated memory including the KV cache is 18.4 GB. The formula is published at https://localllm.az-arte.com/en/guide/memory-calculation.
Can POCKET-26B 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 POCKET-26B run in 8 GB, 12 GB, 16 GB or 24 GB of VRAM?: 8 GB: no quantization fits; 12 GB: up to Q2_K (est. 12.7 GB); 16 GB: up to Q2_K (est. 12.7 GB); 24 GB: up to Q4_K_M (est. 18.4 GB) (estimated at 8,192 tokens with an f16 KV cache, files from the most-downloaded source).
What is the command to run POCKET-26B with Ollama?: Run "ollama run hf.co/FINAL-Bench/POCKET-26B-GGUF:Q4_K_M" to use the GGUF on Hugging Face directly. To use another quantization, replace the name at the end.