Which is better, Gemini 3.6 Flash or Qwen3.5 4B?

Gemini 3.6 Flash vs Qwen3.5 4B

17 SHARED BENCHMARKS

Across 17 shared benchmarks, Gemini 3.6 Flash scores higher on 15 and Qwen3.5 4B on 2. The widest gap is gdpval, where Gemini 3.6 Flash scores 41.6 against 8.3. Qwen3.5 4B is the cheaper of the two on tracked API pricing ($0.03 against $0.75 per million input tokens).

GOOGLEVSALIBABA17 SHARED15–2 HEAD-TO-HEAD

BenchmarkGemini 3.6 FlashMARGINQwen3.5 4B

AA Agentic Index30.236.3

AA Intelligence34.320

AA-LCR8063

AA-Omniscience22.1-58.4

aime_202696.799

Artificial Analysis Coding Index69.222.6

critpt10.60

gdpval41.68.3

GPQA Diamond92.877.1

HLE40.89.9

hmmt_feb_202689.484.1

MMMU-Pro83.265.4

OmniScience Accuracy5015.1

OmniScience Non-Hallucination44.413.4

scicode53.418.3

TauBench V3 - Banking29.96.8

Terminal-Bench 2.17825.8

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. Quoted rates are the price-setter row we currently track for each model — its direct or vendor-official listing where one exists (direct, direct), otherwise the lowest tracked offer.

Compare Gemini 3.6 Flash withALL PAIRINGS →

GoogleGemini 3.1 Pro Gemini 3 Pro Gemma 4 26B A4B Gemini 2.5 Pro Gemini 1.5 Pro Gemma 4 31B

Across the fieldNemotron 3 Super 120B A12B GPT-5.5 Claude Opus 4.8 GPT-4o GPT-5.6 Sol GPT-5.4 GPT-5.2 Claude Opus 4.6

Compare Qwen3.5 4B 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 fieldNemotron 3 Super 120B A12B GPT-5.5 Claude Opus 4.8 GPT-4o GPT-5.6 Sol Gemini 3.1 Pro GPT-5.4 GPT-5.2