👁 Google
Google
Gemma 4 31B
Frontier
Apr 2026Released256K tokensContextApache-2.0License86 StrongQuality
Gemma 4 31B (30.700000762939453B parameters) requires approximately 35.2 GB of VRAM with Q4_K_M quantization. For the best balance of quality and speed, we recommend hardware with at least 41 GB of VRAM.
Get started
— copy & paste to run locallyCopy-paste commands to run Gemma 4 31B on your machine.
Run
ollama run gemma4:31bQuick specs
Parameters30.7B
Architecturedense
Context256K tokens
Modalitytext
Min RAM12 GB
Rec. RAM18.7 GB (Q4_K_M)
LicenseApache-2.0
FamilyGemma
✓ Code✓ Chat✓ Reasoning
About this model
- •Highest quality open Gemma model
- •256K context window
- •2150 Codeforces ELO
- •Apache 2.0 license
- •89.2% AIME 2026
Related models
Your hardware
Detecting...
Quick picks
Best hardware
Top picks for Gemma 4 31B
Run this model
Quantization options
VRAM estimates by quant level
No hardware detected — fit column shows raw VRAM estimates
| Quant | Bits | VRAM | Quality | Fit |
|---|---|---|---|---|
Q2_K | 2 | 12.0 GB | Low | — |
Q3_K_S | 3 | 15.0 GB | Low | — |
NVFP4 | 4 | 17.2 GB | Medium | — |
Q4_K_M | 4 | 18.7 GB | Medium | — |
Q5_K_M | 5 | 22.1 GB | High | — |
Q6_K | 6 | 25.2 GB | High | — |
Q8_0 | 8 | 32.8 GB | Very High | — |
F16 | 16 | 62.9 GB | Maximum | — |
Quality benchmarks
Gemma 4 31B benchmark scores
Coding
SWE-bench Verified—
HumanEval+—
Aider Polyglot—
LiveCodeBench80.0%
Reasoning
MMLU-Pro85.2%
GPQA Diamond84.3%
MATH-500—
ARC Challenge—
Source: official · 2026-04-02
Hardware compatibility
Fit estimates across all hardware
Computing compatibility...
Memory breakdown
Reference: RTX 2060 6GB
Weights18.7 GB
KV Cache14.6 GB
Runtime1.2 GB
Headroom0.6 GB
Frequently asked questions
FAQ — Gemma 4 31B
See also