Gemma 4 26B A4B IT — Hardware Requirements & GPU Compatibility
VisionFunctions
Gemma 4 26B A4B IT applies a mixture-of-experts design to Google's Gemma 4 family, with roughly 26 billion total parameters but only about 4 billion active for any given token. That active-parameter count is what determines inference speed, so despite its total size the model can respond about as quickly as a much smaller dense model, though the full parameter set still needs to be held in memory, putting local use in single high-end consumer GPU territory once quantized. It also accepts image input alongside text, making it usable for multimodal chat and visual question answering. The model offers a 256K token context window, suited to long documents or extended conversations, and is released under the Apache 2.0 license for unrestricted use. Released alongside the dense 31B variant, it gives developers a faster-inference option within the same Gemma 4 generation.
Runs great on a 24 GB GPU
Runs from 12 GB (IQ3_S · 16K)Best on 24 GB (Q4_K_M · ~171 tok/s · 128K)Macs from 32 GB (~26 tok/s)
Score 105.4#37 of 92 open#6 of 34 that fit 24 GBApache 2.0
Reference GPU · RTX 4090 24 GB
Estimates, ±20% · how we estimate
llmrun Composite Score →
105.4
- 90% range 101.7–109.1
- #37 of 92 open models
- #6 of 34 that fit 24 GB
Run Gemma 4 26B A4B IT
Best fit for your MacBook Air 13" M4 (32 GB). Every other method, OS and quant is in the step-by-step guide.
Recommended for your MacBook Air 13" M4 (32 GB)
mlx_lm.chat --model lmstudio-community/gemma-4-26B-A4B-it-MLX-4bitMLX repo with its own quantisation
Download: gemma-4-26B-A4B-it-UD-Q4_K_M.gguf16.95 GBunsloth
Other ways
each opens step-by-stepRun in cloud
Fits on RTX 4000 Ada (20 GB) (3 GB headroom) · Q4_K_M
- Generation speed
- ~75 tok/s
- generation speed
- Cost per 1M output tokens
- $0.74
- per 1M output tokens
How Much VRAM Does Gemma 4 26B A4B IT Need?
Select a quantization level; the listed hardware adheres to it.
| Quantization | VRAM | +32K | On your device | |
|---|---|---|---|---|
| 11.6 GB | 12.5 GB | 2-bit · K-quant — 2-bit quantization with K-quant improvements | Fits: 256K context · ~34 tok/s | |
| 11.9 GB | 12.8 GB | 3-bit small — 3-bit small quantization | Fits: 256K context · ~33 tok/s | |
| 13.2 GB | 14.1 GB | 3-bit medium — 3-bit medium quantization | Fits: 256K context · ~31 tok/s | |
| 13.6 GB | 14.5 GB | 4-bit · legacy — 4-bit legacy quantization | Fits: 256K context · ~30 tok/s | |
| 16.1 GB | 17 GB | 4-bit medium — 4-bit medium quantization — most popular sweet spot | Fits: 256K context · ~26 tok/s | |
| 19.0 GB | 19.9 GB | 5-bit medium — 5-bit medium quantization — good quality/size tradeoff | Fits: 256K context · ~22 tok/s | |
| 21.9 GB | 22.8 GB | 6-bit, very good — 6-bit quantization, very good quality | Fits: 128K context · ~20 tok/s | |
| 26.4 GB | 27.4 GB | 8-bit, near-lossless — 8-bit quantization, near-lossless | Fits: 64K context · ~17 tok/s |
Selected Q4_K_M
4-bit medium quantization16.5GBat 16K context
- Weights
- 15.5 GB
- KV cache
- 0.7 GB
- Overhead
- 0.3 GB
Fits MacBook Air 13" M4 (32 GB) · 12.0 GB free · ~26 tok/s
≥50 tok/s needs ≥230 GB/s · RTX 5090 ~250, RTX 3090 Ti ~171
Which Hardware Can Run Gemma 4 26B A4B IT?
Q4_K_M · 16.1 GBGemma 4 26B A4B IT (Q4_K_M) requires 16.1 GB of VRAM to load the model weights. For comfortable inference with headroom for KV cache and system overhead, 21+ GB is recommended. Using the full 262K context window can add up to 8.7 GB, bringing total usage to 24.8 GB. 8 GPUs can run it, including NVIDIA GeForce RTX 5090, NVIDIA GeForce RTX 3090 Ti. 125 devices with unified memory can run Gemma 4 26B A4B IT, including NVIDIA DGX H100, NVIDIA DGX A100 640GB, Mac Mini M5 Pro (24 GB).
20 GB
fits · 3.5 GB free · up to 64K ctx
1 GPU
RX 7900 XT~85Simulate output on RX 7900 XTfits · 7.5 GB free · up to 128K ctx
6 GPUs · 2 Mac classes
RTX 3090 Ti~171Simulate output on RTX 3090 TiRTX 4090~171Simulate output on RTX 4090RTX 3090~162Simulate output on RTX 3090M4 Pro24 GB~49Simulate output on Mac Mini M4 Pro (24 GB)M424 GB~26Simulate output on iMac 24" M4 (24 GB)48 GB
48 GB tier →fits · 31.5 GB free · up to 256K ctx
2 Mac classes · 1 device family
M4 Max48 GB~76Simulate output on Mac Studio M4 Max (48 GB)M4 Pro48 GB~49Simulate output on Mac Mini M4 Pro (48 GB)Snapdragon X2 Elite Extreme (X2E-96-100)~46Simulate output on Snapdragon X2 Elite Extreme Copilot+ PC
By VRAM tier · typical GPUs and Macs at Q4_K_M
How good is Gemma 4 26B A4B IT?
The llmrun Score merges 9 public benchmark boards into a single figure, where 100 represents the average for open models.
Score vs VRAM in the 24 GB tier
- Gemma 4 26B A4B IT
- Other models
- Best Score for its VRAM
#6 of 34 that fit 24 GBEvery model here fits your MacBook Air 13" M4 (32 GB) (32 GB)
Each dot is a model at the best quant that fits a 24 GB GPU. The line joins the models with the best Score for their VRAM; a dot below it is beaten by a model that needs no more memory.
Full 24 GB ranking →Benchmarks
The 6 boards that weigh most in the llmrun Score, of 11 with a result for this model; badges show who ran the test.
- GPQA Diamond
Epoch AI · Apr 2026Independent
Independent73.2%
#29/97 open
- LMArena Text
LMArena · Oct 2026Arena
Arena1437.6
#23/186 open
- AIME 2024/2025
Epoch AI · Apr 2026Independent
Independent82.2%
#23/85 open
- LMArena WebDev
LMArena · Oct 2026Arena
Arena1359.0
#33/50 open
- SciCode
Epoch AI · Apr 2026Lab-reported
Lab-reported40.1%
#30/65 open
- WeirdML
Epoch AI · Apr 2026Independent
Independent35.2%
#34/46 open
In the 24 GB class
34 models fit 24 GBSame Score, only models that fit 24 GB at their best quant: the ranking a 24 GB owner actually chooses from.
- #1
Qwen3.8 27BAlibaba
121.5
~38 tok/s · 17.9 GB
- #2
Qwen3.6 27BAlibaba
111.4
~38 tok/s · 17.9 GB
- #3
Qwen3.6 35B A3BAlibaba
106.9
~151 tok/s · 22.0 GB
- #4
Gemma 4 31B ITGoogle
106.1
~32 tok/s · 21.7 GB
- #5
Qwen3.5 35B A3BAlibaba
105.6
~151 tok/s · 22.0 GB
Alternatives to Gemma 4 26B A4B IT
Gemma 4 family · 35 models →Official models with an llmrun Score, each answering one question; numbers are at the best quant that fits a 24 GB GPU.
Best alternative at this size
Qwen3.8 27B
Alibaba
- 16.1 higher Score
- 1.4 GB more VRAM
- 133 tok/s slower
Score 121.5 · 17.9 GB · ~38 tok/s
Fits in less VRAM
GPT OSS 20B
OpenAI
- 9.2 lower Score
- 3.3 GB less VRAM
- 19 tok/s faster
Score 96.2 · 13.1 GB · ~190 tok/s
Other vendor
Qwen3.6 27B
Alibaba
- 6 higher Score
- 1.4 GB more VRAM
- 133 tok/s slower
Score 111.4 · 17.9 GB · ~38 tok/s
Similar models
10| Model | Score | VRAM | tok/s | Why it is similar |
|---|---|---|---|---|
| Qwen3.6 35B A3BAlibaba | 106.9 | 22 GB | ~151 | similar score |
| Gemma 4 31B ITGoogle | 106.1 | 21.7 GB | ~32 | similar score |
| Qwen3.5 35B A3BAlibaba | 105.6 | 22 GB | ~151 | similar score |
| Qwen3.5 27BAlibaba | 103.8 | 17.9 GB | ~38 | similar score |
| Seed OSS 36B InstructByteDance-Seed | 102.5 | 22.8 GB | ~31 | similar score |
| Qwen3 30B A3B Thinking 2507Alibaba | 100.6 | 19.4 GB | ~173 | similar score |
| Qwen3.5 9BAlibaba | 99.4 | 6.6 GB | ~103 | similar score |
| Qwen3 32BAlibaba | 98.4 | 22.6 GB | ~32 | similar score |
| Qwen3 14BAlibaba | 97.9 | 11.9 GB | ~69 | similar score |
| DeepSeek R1 Distill Qwen 32BDeepSeek | 97.0 | 23 GB | ~34 | similar score |
Specifications
- Publisher
- Family
- Gemma 4
- Parameters
- 25.8B
- Active parameters
- 4.4B per token
- Architecture
- Gemma4ForConditionalGeneration
- Context Length
- 262,144 tokens
- Vocabulary Size
- 262,144
- Release Date
- 2026-03-11
- License
- Apache 2.0
Frequently Asked Questions
How much VRAM does Gemma 4 26B A4B IT need?
Gemma 4 26B A4B IT requires 16.1 GB of VRAM at Q4_K_M, or 52.3 GB at BF16. Full 262K context adds up to 8.7 GB (24.8 GB total).
VRAM = Weights + KV Cache + Overhead
Weights = 25.8B × 4.8 bits ÷ 8 = 15.5 GB
KV Cache + Overhead ≈ 0.6 GB (at 2K context + ~0.3 GB framework)
Fit ratings and hardware model lists check this model with room for a 16K-token context, which needs a little more memory.
KV Cache + Overhead ≈ 9.3 GB (at full 262K context)
VRAM usage by quantization
Can NVIDIA GeForce RTX 4090 run Gemma 4 26B A4B IT?
Yes, at Q6_K (21.9 GB) or lower. Higher quantizations like Q8_0 (26.4 GB) exceed the NVIDIA GeForce RTX 4090's 24 GB.
What's the best quantization for Gemma 4 26B A4B IT?
For Gemma 4 26B A4B IT, Q4_K_M (16.1 GB) offers the best balance of quality and VRAM usage. Q4_K_L (16.4 GB) provides better quality if you have the VRAM. The smallest option is IQ2_XXS at 7.7 GB.
VRAM requirement by quantization
★ Recommended — best balance of quality and VRAM usage.
Can I run Gemma 4 26B A4B IT on a Mac?
Yes, but only at lower quantizations. The smallest Mac that can run Gemma 4 26B A4B IT is iPad Pro M5 13" (16 GB) at IQ2_XXS; 136 of the 156 Macs we list can run it at some quantization. For Q4_K_M (16.1 GB) you need a Mac with more unified memory.
Can I run Gemma 4 26B A4B IT locally?
Yes — Gemma 4 26B A4B IT can run locally on consumer hardware. Popular tools include Ollama, LM Studio, and llama.cpp.
How fast is Gemma 4 26B A4B IT?
At Q4_K_M, Gemma 4 26B A4B IT can reach ~152 tok/s on AMD Instinct MI350X. On NVIDIA GeForce RTX 4090: ~171 tok/s. Speed depends mainly on GPU memory bandwidth. Real-world results typically within ±20%.
tok/s = 1000 ÷ (active GB ÷ (bandwidth GB/s × efficiency) × 1000 + layers × routing ms)
Mixture-of-Experts: just the active experts get read for each token, along with a fixed routing cost per layer.
Example: NVIDIA B200 → 2.7 GB active ÷ (8000 × 0.65) = 0.53 ms, plus 30 layers × 0.055 ms = 1.65 ms, so 1000 ÷ 2.18 ms = ~459 tok/s
Estimated speed at Q4_K_M (16.1 GB)
Real-world results typically within ±20%. Speed depends on batch size, quantization kernel, and software stack.
What's the download size of Gemma 4 26B A4B IT?
At Q4_K_M, the download is about 15.48 GB. The full-precision BF16 version is 51.61 GB. The smallest option (IQ2_XXS) is 7.10 GB.
Which GPUs can run Gemma 4 26B A4B IT?
8 consumer GPUs can run Gemma 4 26B A4B IT at Q4_K_M. Top options include NVIDIA GeForce RTX 5090, AMD Radeon RX 7900 XT, AMD Radeon RX 7900 XTX. 1 GPU have plenty of headroom for comfortable inference.
Which devices can run Gemma 4 26B A4B IT?
125 devices with unified memory can run Gemma 4 26B A4B IT at Q4_K_M (16.1 GB), including AMD Ryzen AI 9 HX 370 (Strix Point) Laptop, ASUS Ascent GX10, Asus ROG Flow Z13 (2025, Ryzen AI Max+ 395, 128 GB), Beelink GTR9 Pro (Ryzen AI Max+ 395, 128 GB). Apple Silicon Macs use unified memory shared between CPU and GPU, making them well-suited for local LLM inference.