Skip to content

Gemma 4 31B

by Google · gemma-4 family

31B

parameters

text-generation code-generation reasoning multilingual vision tool-use math creative-writing summarization

Gemma 4 31B is the flagship of the Gemma 4 family — a 30.7B dense model that ranks #3 on Arena AI among all open models, outcompeting models with 20x more parameters. It scores 84.3% on GPQA Diamond, 89.2% on AIME 2026, and 80.0% on LiveCodeBench. At Q4 it needs about 22 GB VRAM, fitting on a RTX 3090/4090/5090 or a Mac with 24 GB+ unified memory. Released under Apache 2.0, it's one of the most permissively licensed frontier-class open models available.

Quick Start with Ollama

ollama run 31b-it-q4_K_M
Resources Ollama Hugging Face Official Page Research Paper
Creator Google
Parameters 31B
Architecture transformer-decoder
Context 256K tokens
Released Apr 2, 2026
License Apache 2.0
Ollama gemma4:31b

Quantization Options

Format File Size VRAM Required Quality Ollama Tag
Q4_K_M rec 20 GB 22 GB 31b-it-q4_K_M
Q8_0 34 GB 38 GB 31b-it-q8_0
F16 62 GB 66 GB 31b-it-fp16

Compatible Hardware

Q4_K_M requires 22 GB VRAM

Compatible Hardware

HardwareVRAMTypeFitEst. Speed
Mac Studio M4 Ultra 512GB512 GBmacRuns~37 tok/s
Mac Pro M2 Ultra 192GB192 GBmacRuns~36 tok/s
Mac Studio M4 Ultra 192GB192 GBmacRuns~37 tok/s
Mac Studio M4 Max 128GB128 GBmacRuns~25 tok/s
MacBook Pro M4 Max 128GB128 GBmacRuns~25 tok/s
MacBook Pro M5 Max 128GB128 GBmacRuns~25 tok/s
NVIDIA RTX PRO 6000 Blackwell96 GBgpuRuns~87 tok/s
MacBook Pro M3 Max 96GB96 GBmacRuns~18 tok/s
Mac mini M4 Pro 64GB64 GBmacRuns~12 tok/s
Mac Studio M4 Max 64GB64 GBmacRuns~25 tok/s
MacBook Pro M4 Max 64GB64 GBmacRuns~25 tok/s
MacBook Pro M5 Max 64GB64 GBmacRuns~25 tok/s
NVIDIA RTX 6000 Ada Generation48 GBgpuRuns~44 tok/s
NVIDIA RTX A600048 GBgpuRuns~35 tok/s
NVIDIA RTX PRO 5000 Blackwell48 GBgpuRuns~44 tok/s
Mac mini M4 Pro 48GB48 GBmacRuns~12 tok/s
MacBook Pro M3 Max 48GB48 GBmacRuns~18 tok/s
MacBook Pro M4 Max 48GB48 GBmacRuns~25 tok/s
MacBook Pro M4 Pro 48GB48 GBmacRuns~12 tok/s
MacBook Pro M5 Max 48GB48 GBmacRuns~19 tok/s
MacBook Pro M5 Pro 48GB48 GBmacRuns~12 tok/s
Mac Studio M4 Max 36GB36 GBmacRuns~25 tok/s
MacBook Pro M3 Pro 36GB36 GBmacRuns~7 tok/s
MacBook Pro M5 Max 36GB36 GBmacRuns~19 tok/s
NVIDIA RTX 5000 Ada Generation32 GBgpuRuns~33 tok/s
NVIDIA GeForce RTX 509032 GBgpuRuns~81 tok/s
iMac M4 32GB32 GBmacRuns~5 tok/s
Mac mini M4 32GB32 GBmacRuns~5 tok/s
MacBook Air M4 32GB32 GBmacRuns~5 tok/s
MacBook Air M5 32GB32 GBmacRuns~5 tok/s
MacBook Pro M5 32GB32 GBmacRuns~5 tok/s
AMD Radeon RX 7900 XTX24 GBgpuRuns (tight)~44 tok/s
NVIDIA GeForce RTX 309024 GBgpuRuns (tight)~43 tok/s
NVIDIA GeForce RTX 3090 Ti24 GBgpuRuns (tight)~46 tok/s
NVIDIA GeForce RTX 409024 GBgpuRuns (tight)~46 tok/s
NVIDIA RTX A500024 GBgpuRuns (tight)~35 tok/s
iMac M3 24GB24 GBmacRuns (tight)~5 tok/s
Mac mini M2 24GB24 GBmacRuns (tight)~5 tok/s
Mac mini M4 Pro 24GB24 GBmacRuns (tight)~12 tok/s
MacBook Air M2 24GB24 GBmacRuns (tight)~5 tok/s
MacBook Air M4 24GB24 GBmacRuns (tight)~5 tok/s
MacBook Air M5 24GB24 GBmacRuns (tight)~5 tok/s
MacBook Pro M4 Pro 24GB24 GBmacRuns (tight)~12 tok/s
MacBook Pro M5 24GB24 GBmacRuns (tight)~5 tok/s
MacBook Pro M5 Pro 24GB24 GBmacRuns (tight)~12 tok/s
AMD Radeon RX 7900 XT20 GBgpuCPU Offload~11 tok/s
NVIDIA RTX 4000 Ada Generation20 GBgpuCPU Offload~5 tok/s
MacBook Pro M3 Pro 18GB18 GBmacCPU Offload~2 tok/s
AMD Radeon RX 6900 XT16 GBgpuCPU Offload~7 tok/s
AMD Radeon RX 6800 XT16 GBgpuCPU Offload~7 tok/s
AMD Radeon RX 7800 XT16 GBgpuCPU Offload~8 tok/s
AMD Radeon RX 9060 XT 16GB16 GBgpuCPU Offload~7 tok/s
AMD Radeon RX 9070 XT16 GBgpuCPU Offload~9 tok/s
AMD Radeon RX 907016 GBgpuCPU Offload~7 tok/s
Intel Arc A77016 GBgpuCPU Offload~8 tok/s
NVIDIA GeForce RTX 4060 Ti 16GB16 GBgpuCPU Offload~4 tok/s
NVIDIA GeForce RTX 4070 Ti Super16 GBgpuCPU Offload~9 tok/s
NVIDIA GeForce RTX 4080 Super16 GBgpuCPU Offload~10 tok/s
NVIDIA GeForce RTX 408016 GBgpuCPU Offload~10 tok/s
NVIDIA GeForce RTX 5060 Ti 16GB16 GBgpuCPU Offload~6 tok/s
NVIDIA GeForce RTX 5070 Ti16 GBgpuCPU Offload~12 tok/s
NVIDIA GeForce RTX 508016 GBgpuCPU Offload~13 tok/s
NVIDIA RTX A400016 GBgpuCPU Offload~6 tok/s
iMac M1 16GB16 GBmacCPU Offload~1 tok/s
iMac M4 16GB16 GBmacCPU Offload~2 tok/s
Mac mini M1 16GB16 GBmacCPU Offload~1 tok/s
Mac mini M4 16GB16 GBmacCPU Offload~2 tok/s
MacBook Air M2 16GB16 GBmacCPU Offload~2 tok/s
MacBook Air M3 16GB16 GBmacCPU Offload~2 tok/s
MacBook Air M4 16GB16 GBmacCPU Offload~2 tok/s
MacBook Air M5 16GB16 GBmacCPU Offload~2 tok/s
MacBook Pro M1 16GB16 GBmacCPU Offload~1 tok/s
MacBook Pro M2 Pro 16GB16 GBmacCPU Offload~3 tok/s
MacBook Pro M5 16GB16 GBmacCPU Offload~2 tok/s
33 hardware device(s) cannot run this model at Q4_K_M.

Benchmark Scores

85.2
mmlu_pro
84.3
gpqa_diamond
89.2
aime2026
80.0
livecodebench