~/model/huggingfacegoogle/gemma-4-26B-A4B-it
Google provider

Gemma 4 26B A4B

Google's 2026 MoE: 26B total, 4B active. Vision-ready.

Parameters
26B
Family
Google
Context
128K tokens
FP16 weights
52GB
// where you can run it
OllamaLM StudiovLLMMLXoMLX
// hugging face stats (cached daily)
9.4M downloads · 981 likes · license: apache-2.0 · updated 3 days ago

What you need to run this.

$ ./vrambudget --model gemma-4-26b --by quant
FP1652 GB
H100 80GB80 GB · fitsM3 Max 9696 GB · fitsRTX Pro 600096 GB · fitsM4 Max 128128 GB · fits
Q8_028 GB
RTX A600048 GB · fitsRTX 6000 Ada48 GB · fitsL40S48 GB · fitsRadeon Pro W790048 GB · fits
Q6_K21 GB
RTX 509032 GB · fitsRTX A600048 GB · fitsRTX 6000 Ada48 GB · fitsL40S48 GB · fits
Q5_K_M18 GB
RTX 309024 GB · fitsRTX 3090 Ti24 GB · fitsRTX 409024 GB · fitsRX 7900 XTX24 GB · fits
Q4_K_M15 GB
RTX 309024 GB · fitsRTX 3090 Ti24 GB · fitsRTX 409024 GB · fitsRX 7900 XTX24 GB · fits

// budgets shown at ctx 8K, concurrency 1, 15% safety headroom. Tune in the calculator →

Alternatives at this size.

$ grep --params similar catalog.json
$ similarQwen 3.6 27B27BAlibaba$ similarMistral Small 324BMistral$ similarQwen3 30B A3B30.5BAlibaba$ similargpt-oss 20B20.9BOpenAI

Discussion.

$ gh discussion list

// sign in with github to leave a comment. threads live in the repo's discussions tab.