RunsOnMy ModelsGPUs & MacsBy VRAMAPI pricesVRAM calculatorNew

Can I run gemma 4 26B A4B it locally?

gemma 4 26B A4B it by google has 26B parameters (4.0B active per token, 128 experts, 8 used per token). At 4-bit with an 8K context it needs about 17.2 GB of GPU memory. It fits on 58 of 65 GPUs and Macs we track. See all models that fit in 24 GB of VRAM.

17.2 GBQ4_K_M · 8K context
RTX 3090smallest consumer GPU that fits
M4 32GBsmallest Mac that fits
256Kmax context
GQA + sliding windowattention type

Memory needed by quant and context length

GB of GPU memory = weights + KV cache (FP16) + runtime overhead. Weights use real file sizes from unsloth/gemma-4-26B-A4B-it-GGUF where available (marked •), otherwise parameter count × bits per weight. How we calculate.

QuantQuality4K ctx8K ctx32K ctx128K ctx
2-bit (Q2_K)rough10.3 GB10.4 GB11.4 GB15.1 GB
3-bit (Q3_K_M) •usable13.1 GB13.3 GB14.2 GB17.9 GB
4-bit (Q4_K_M) •good, most popular17.0 GB17.2 GB18.1 GB21.9 GB
5-bit (Q5_K_M) •very good20.9 GB21.1 GB22.0 GB25.8 GB
6-bit (Q6_K) •near lossless22.8 GB23.0 GB23.9 GB27.7 GB
8-bit (Q8_0) •effectively lossless26.3 GB26.4 GB27.4 GB31.1 GB
16-bit (F16/BF16) •original weights48.3 GB48.4 GB49.4 GB53.1 GB

Which GPUs and Macs run gemma 4 26B A4B it?

Best quality quant that fits at 8K context. "Tight" = within 10% of usable memory. Speed is a ceiling from memory bandwidth (real speed is typically 50–80% of it).

HardwareMemoryFits?Best quantNeedsMax tok/s (est.)
RTX 3060 12GB12 GBTightQ2_K10.4 GB240details
RTX 4060 Ti 16GB16 GBFitsQ3_K_M13.3 GB146details
RTX 309024 GBFitsQ4_K_M17.2 GB356details
RTX 409024 GBFitsQ4_K_M17.2 GB384details
RTX 509032 GBFitsQ8_026.4 GB430details
RTX 5070 Ti16 GBFitsQ3_K_M13.3 GB454details
RX 7900 XTX24 GBFitsQ4_K_M17.2 GB365details
Mac M4 Pro 48GB48 GBFitsQ8_026.4 GB66details
Mac M4 Max 128GB128 GBFitsF1648.4 GB70details
Mac M3 Ultra 512GB512 GBFitsF1648.4 GB105details
DGX Spark128 GBFitsF1648.4 GB35details
RTX 407012 GBTightQ2_K10.4 GB336details
RTX 5060 Ti 16GB16 GBFitsQ3_K_M13.3 GB227details
RTX 40608 GBNo———details
Arc B58012 GBTightQ2_K10.4 GB304details
RTX 3060 Ti8 GBNo———details
RTX 30708 GBNo———details
RTX 3080 10GB10 GBNo———details
RTX 3080 12GB12 GBTightQ2_K10.4 GB608details
RTX 3090 Ti24 GBFitsQ4_K_M17.2 GB384details
RTX 4060 Ti 8GB8 GBNo———details
RTX 4070 Super12 GBTightQ2_K10.4 GB336details
RTX 4070 Ti12 GBTightQ2_K10.4 GB336details
RTX 4070 Ti Super16 GBFitsQ3_K_M13.3 GB341details
RTX 408016 GBFitsQ3_K_M13.3 GB363details
RTX 4080 Super16 GBFitsQ3_K_M13.3 GB373details
RTX 50608 GBNo———details
RTX 507012 GBTightQ2_K10.4 GB448details
RTX 508016 GBFitsQ3_K_M13.3 GB487details
2x RTX 30902×24 GBFitsQ8_026.4 GB225details
2x RTX 40902×24 GBFitsQ8_026.4 GB242details
2x RTX 50902×32 GBFitsF1648.4 GB229details
RTX A600048 GBFitsQ8_026.4 GB184details
RTX 6000 Ada48 GBFitsQ8_026.4 GB231details
RTX PRO 6000 Blackwell96 GBFitsF1648.4 GB229details
L40S48 GBFitsQ8_026.4 GB208details
A100 40GB40 GBFitsQ8_026.4 GB373details
A100 80GB80 GBFitsF1648.4 GB260details
H100 PCIe80 GBFitsF1648.4 GB255details
H100 SXM80 GBFitsF1648.4 GB428details
H200141 GBFitsF1648.4 GB613details
RX 76008 GBNo———details
RX 7700 XT12 GBTightQ2_K10.4 GB288details
RX 7800 XT16 GBFitsQ3_K_M13.3 GB316details
RX 7900 GRE16 GBFitsQ3_K_M13.3 GB292details
RX 7900 XT20 GBFitsQ3_K_M13.3 GB405details
RX 907016 GBFitsQ3_K_M13.3 GB324details
RX 9070 XT16 GBFitsQ3_K_M13.3 GB324details
Ryzen AI Max+ 395 128GB128 GBFitsF1648.4 GB33details
Arc A770 16GB16 GBFitsQ3_K_M13.3 GB284details
Mac M1 16GB16 GBTightQ2_K10.4 GB45details
Mac M2 24GB24 GBFitsQ3_K_M13.3 GB51details
Mac M3 24GB24 GBFitsQ3_K_M13.3 GB51details
Mac M4 16GB16 GBTightQ2_K10.4 GB80details
Mac M4 32GB32 GBFitsQ4_K_M17.2 GB46details
Mac M1 Pro 32GB32 GBFitsQ4_K_M17.2 GB76details
Mac M3 Pro 36GB36 GBFitsQ5_K_M21.1 GB46details
Mac M4 Pro 24GB24 GBFitsQ3_K_M13.3 GB138details
Mac M4 Pro 64GB64 GBFitsQ8_026.4 GB66details
Mac M1 Max 64GB64 GBFitsQ8_026.4 GB96details
Mac M2 Max 96GB96 GBFitsF1648.4 GB51details
Mac M3 Max 128GB128 GBFitsF1648.4 GB51details
Mac M4 Max 64GB64 GBFitsQ8_026.4 GB131details
Mac M2 Ultra 192GB192 GBFitsF1648.4 GB102details
Mac M3 Ultra 256GB256 GBFitsF1648.4 GB105details

Prefer an API?

Google: Gemma 4 26B A4B is on OpenRouter at $0.0765 per 1M input tokens and $0.255 per 1M output tokens. Price details and history →

Questions

How much VRAM does gemma 4 26B A4B it need?

About 17.2 GB at 4-bit (Q4_K_M) with an 8K context: 15.8 GB of weights (real GGUF file size), 0.5 GB of KV cache and 0.9 GB of runtime overhead. At 8-bit it needs about 26.4 GB.

What is the smallest GPU that runs gemma 4 26B A4B it?

At Q4_K_M and 8K context, the smallest single consumer GPU in our table that fits it is the RTX 3090 (24 GB).

Can I run gemma 4 26B A4B it on a Mac?

Yes. At Q4_K_M with 8K context it fits on a Mac M4 32GB or larger, using the default macOS GPU memory limit.

Similar models

Same family or a similar parameter count, with the VRAM each needs at Q4_K_M and 8K context.

diffusiongemma 26B A4B it

26B · 17.1 GB at 4-bit

gemma 3 27b it

27B · 17.3 GB at 4-bit

gemma 4 31B it

31B · 20.0 GB at 4-bit

GLiDE 27B

26B · 16.2 GB at 4-bit

Trinity Mini

26B · 15.8 GB at 4-bit

Arete 27B

27B · 17.4 GB at 4-bit

Model facts

Hugging Face repogoogle/gemma-4-26B-A4B-it
Parameters25,805,936,206 (26B) · ~4.0B active (from the model name)
Layers / hidden size30 / 2816
AttentionGQA + sliding window; 16 heads, 8 KV heads, head dim 256; sliding window 1024
KV cache per token (FP16)240.0 KiB (sliding layers capped at the window)
Max context262,144 tokens
Licenseapache-2.0
Released on HFMar 11, 2026
Downloads (30d) / likes5,102,751 / 1,780

Sources: Hugging Face model API · config.json (google/gemma-4-26B-A4B-it) · GGUF files · GPU specs from vendor pages (list). Data fetched Nov 17, 2026. Spot an error? Report it.