OpenModelsHQ

gemma-4-26B-A4B-it

GoogleVisionMoEApache-2.0

gemma-4-26B-A4B-it is an open-weight vision and language model from Google with 25.8B parameters (mixture of experts). At 4-bit it needs about 17.4 GB of memory, so the smallest current device that fits it is the Arc Pro B60.

Overview

Estimated memory at 4-bit
17.4GB

Includes room for an 8k context and runtime overhead. More context needs more memory.

Total parameters
25.8B
Mixture of experts, all weights counted
Context window
256K
262,144 tokens
Cheapest API, blended / 1M
$0.0208
₹1.96 · PrimeIntellect · 3:1 input to output
All Google API prices

Key information

Lab
Google
Released
11 Mar 2026
Parameters
25.8B · MoE
Type
Vision and language
Context
262,144 tokens
License
Apache-2.0
HF downloads, 30 days
5,251,952
View all

Will it run on your hardware?

How we estimate
Memory needed vs devices
gemma-4-26B-A4B-it needs about 17.4 GB at 4-bit, 30.5 GB at 8-bit and 57.1 GB at 16-bit. At 4-bit it fits GeForce RTX 5090, DGX Spark, Mac Studio M5 Ultra (256GB). Dashed lines are usable memory on each device; a model fits when its estimate stays within 90% of it. Log scale. 8k context.
Memory needed vs devices: data
Estimate or deviceMemory
Q417.4 GB
Q830.5 GB
BF1657.1 GB
RTX 5060 Ti 16 GB16 GB usable
RTX 5090 32 GB32 GB usable
DGX Spark 128 GB108.8 GB usable
Mac Studio M5 Ultra 256 GB192 GB usable

Single-device memory estimate. Pick a device to update the On your device column.

36 GB usable memory

FormatMemory neededOn your deviceExample hardware
4-bit (Q4)≈ 17.4 GBFits 48%Arc Pro B6024 GB memory
8-bit (Q8)≈ 30.5 GBFits 85%L40S48 GB memory
16-bit (BF16)≈ 57.1 GBToo big 159%A100 80GB SXM80 GB memory

Fits means the estimate stays within 90% of usable memory. Tight is above 90% and up to 100%. Too big is above 100%. These are memory checks, not speed or compatibility tests.

Mac allowances are 66% of total RAM up to 36 GB, and 75% above. AI boxes allow 85% because their memory is shared with the operating system.

Where to run it

On a Mac

Start with the MLX files, built for Apple Silicon.

lmstudio-community/gemma-4-26B-A4B-it-MLX-4bitCheck your Mac

With GGUF

Choose a file for llama.cpp or a compatible app.

unsloth/gemma-4-26B-A4B-it-GGUFFind compatible hardware

Through an API

PrimeIntellect (bf16) is the cheapest provider we found.

$0.014 / $0.041

Input / output per 1M tokens
₹1.32 / ₹3.87 at ₹94.29 per $1

Blended: $0.0208 (₹1.96) per 1M. Ranked by a 3:1 mix of input and output tokens.

MLX on Apple SiliconView repository
mlx_vlm.generate --model lmstudio-community/gemma-4-26B-A4B-it-MLX-4bit \
  --prompt "Describe this image." --image image.jpg
GGUF with llama.cppView repository
llama-cli -hf unsloth/gemma-4-26B-A4B-it-GGUF \
  -c 8192 -p "Explain how a transformer works."

Example commands built from the repository ids above. They need a working install of MLX or llama.cpp. We have not run them; check each repository for available files.

Hosted API 15

USD and INR per million tokensAll Google API prices
Precision
PrimeIntellectCheapest$0.0208₹1.96$0.014₹1.32$0.041₹3.87262,144BF16
Darkbloom$0.0675₹6.36$0.04₹3.77$0.15₹14131,072Not listed
NextBit$0.1069₹10$0.0675₹6.36$0.225₹21262,144BF16
AkashML$0.1125₹11$0.07₹6.60$0.24₹23131,072FP4
Io Net$0.1227₹12$0.067₹6.32$0.29₹27262,142BF16
DekaLLM$0.1275₹12$0.06₹5.66$0.33₹31262,144BF16
DeepInfra$0.1375₹13$0.07₹6.60$0.34₹32262,144FP8
Makora$0.1375₹13$0.09₹8.49$0.28₹26256,000Not listed
Cloudflare$0.15₹14$0.10₹9.43$0.30₹28256,000Not listed
CoreWeave$0.15₹14$0.10₹9.43$0.30₹28262,144BF16
Novita$0.1975₹19$0.13₹12$0.40₹38262,144BF16
Parasail$0.1975₹19$0.13₹12$0.40₹38262,144BF16
Venice$0.1975₹19$0.13₹12$0.40₹38256,000BF16
SiliconFlow$0.205₹19$0.14₹13$0.40₹38262,144FP8
Google$0.2625₹25$0.15₹14$0.60₹57262,144Not listed
View on OpenRouterEndpoint sourceSorted by the blended price: 3 parts input to 1 part output. Context limits differ by provider.

INR at ₹94.29 per $1. Prices fetched 14 Nov 2026.

More about this model

Frequently asked questions

How much memory does gemma-4-26B-A4B-it need? At an 8k context, about 17.4 GB at 4-bit, 30.5 GB at 8-bit and 57.1 GB at 16-bit (BF16). A longer context needs more.

Can I run gemma-4-26B-A4B-it on a Mac? The smallest current Mac that fits the 4-bit estimate of 17.4 GB is the Mac mini M6 (32GB). An MLX build is available: lmstudio-community/gemma-4-26B-A4B-it-MLX-4bit.

Does the full context fit? gemma-4-26B-A4B-it supports up to 262,144 tokens of context. Our memory estimates use 8k, so long prompts need more memory than shown.

What does it cost through an API? The cheapest provider we found is PrimeIntellect, at $0.01 per 1M input tokens and $0.04 per 1M output tokens (₹1.32 and ₹3.87 at 94.29 rupees per dollar). Ranked by a 3:1 mix of input and output tokens.

License in plain English

Apache-2.0. This license permits commercial use, modification and redistribution, with conditions such as keeping the required notices. This summary is not legal advice.

Read Apache-2.0

Read about this license

OMHQ RigBench

Not yet tested. We show no speed or quality score without a measurement.

Revision history and sources