Which is better, Gemma 4 31B or Qwen3.5 2B?

Gemma 4 31B vs Qwen3.5 2B

31 SHARED BENCHMARKS

Across 31 shared benchmarks, Gemma 4 31B scores higher on 28 and Qwen3.5 2B on 2, with 1 level. The widest gap is Terminal-Bench Hard, where Gemma 4 31B scores 36.4 against 3.8.

GOOGLEVSALIBABA31 SHARED28–2 HEAD-TO-HEAD

BenchmarkGemma 4 31BMARGINQwen3.5 2B

AA Agentic Index11.10.8

AA Intelligence296.9

AA-LCR69.725.6

AA-Omniscience-47.9-60.2

Artificial Analysis Coding Index43.42.9

C-Eval82.673.2

critpt1.40

gdpval12.80

GPQA Diamond85.751.6

HallusionBench67.465.5

HLE26.55

IFBench75.641.3

MathVista79.348.7

mmlu_redux93.779.6

MMLU-Pro85.266.5

mmmlu88.463.1

MMMU-Pro76.943

MMStar77.355.1

OCRBench86.186.1

OmniScience Accuracy208.1

OmniScience Non-Hallucination18.125.8

RealWorldQA72.365.1

scicode45.57.2

SimpleVQA52.935.2

supergpqa65.737.5

t2-bench86.448.8

TauBench V3 - Banking14.82.5

Terminal-Bench 2.143.43

Terminal-Bench Hard36.43.8

τ²-Bench Telecom (AA run)65.581.6

τ³-Bench67.56.2

Best tracked score per model per benchmark (default configuration; source-attributed). ↓ marks lower-is-better metrics. Open either model for its full surface, provenance and pricing.

Compare Gemma 4 31B withALL PAIRINGS →

GoogleGemini 3.1 Pro Gemini 3 Pro Gemma 4 26B A4B Gemini 2.5 Pro Gemini 1.5 Pro Gemma 4 E2B

Across the fieldQwen3.6 27B Qwen3.7 Max Qwen3.8 Max Preview Qwen3.7 Plus Preview Claude Opus 4 Qwen3.5 4B Gemini 3.6 Flash Nemotron 3 Super 120B A12B

Compare Qwen3.5 2B withALL PAIRINGS →

AlibabaQwen3.5 35B A3B Qwen3 VL 4B (Reasoning)Qwen3.5 397B A17B Qwen2.5 Instruct 72B Qwen3.5 27B Qwen3.6 Plus

Across the fieldKimi K2 Instruct Gemma 4 E2B GPT-4o mini Qwen3.6 35B A3B Gemini 3.5 Flash GPT-6 Astra MiniMax M2.7 Qwen3.6 27B