Which is better, Qwen3.5 2B or Qwen3.5 397B A17B?

Qwen3.5 2B vs Qwen3.5 397B A17B

35 SHARED BENCHMARKS

Across 35 shared benchmarks, Qwen3.5 2B scores higher on 1 and Qwen3.5 397B A17B on 34. The widest gap is scicode, where Qwen3.5 397B A17B scores 44.8 against 7.2.

ALIBABAVSALIBABA35 SHARED1–34 HEAD-TO-HEAD

BenchmarkQwen3.5 2BMARGINQwen3.5 397B A17B

AA Agentic Index0.853.3

AA Intelligence6.933

AA-LCR25.677.3

AA-Omniscience-60.2-30.8

Artificial Analysis Coding Index2.948.2

bfcl_v443.672.9

C-Eval73.293

critpt01.7

gdpval035.8

Global-MMLU-Lite64.590

GPQA Diamond51.689.3

HLE529

IFBench41.378.8

ifeval78.692.6

include55.485.6

longbench_v238.763.2

mmlu_prox52.384.7

mmlu_redux79.694.9

MMLU-Pro66.587.8

mmmlu63.188.5

MMMU-Pro4379

MMStar55.183.8

multichallenge33.767.6

OmniScience Accuracy8.130.8

OmniScience Non-Hallucination25.817.3

RealWorldQA65.183.9

scicode7.244.8

SimpleVQA35.267.1

supergpqa37.570.4

t2-bench48.886.7

TauBench V3 - Banking2.513.4

Terminal-Bench 2.1351.3

Terminal-Bench Hard3.840.9

τ²-Bench Telecom (AA run)81.695.6

τ³-Bench6.213.4

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 Qwen3.5 2B withALL PAIRINGS →

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

Across the fieldClaude Opus 4.7 Gemini 2.5 Pro Gemini 1.5 Pro Claude Sonnet 4 Claude Sonnet 5 Muse Spark 1.1 GLM-5.1 GLM-5.2

Compare Qwen3.5 397B A17B withALL PAIRINGS →

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

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