Refresh Models —Deals —Regulatory —Sources —

GPT-5.2 vs Qwen3.5 2B

25 SHARED BENCHMARKS

Across 25 shared benchmarks, GPT-5.2 scores higher on 24 and Qwen3.5 2B on 1. The widest gap is HLE, where GPT-5.2 scores 45.5 against 5.

OPENAIVSALIBABA25 SHARED24–1 HEAD-TO-HEAD

BenchmarkGPT-5.2MARGINQwen3.5 2B

AA Agentic Index60.20.8

AA Intelligence426.9

AA-LCR82.725.6

AA-Omniscience0.3-60.2

Artificial Analysis Coding Index48.72.9

GPQA Diamond92.451.6

HLE45.55

IFBench75.441.3

longbench_v254.538.7

MathVista82.848.7

mmmlu89.663.1

OCRBench80.786.1

OmniScience Accuracy44.38.1

OmniScience Non-Hallucination38.425.8

scicode52.17.2

SimpleVQA55.835.2

TauBench V3 - Banking11.12.5

Terminal-Bench Hard62.23.8

τ²-Bench Telecom (AA run)98.781.6

τ³-Bench126.2

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 GPT-5.2 withALL PAIRINGS →

OpenAIGPT-5.5 GPT-4o GPT-5.6 Sol GPT-5.4 GPT-5 o3

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 fieldClaude Opus 4.6 Claude Sonnet 4.5 Claude 3.5 Sonnet Gemini 3 Pro Claude Fable 5 GPT-5 Claude Opus 4.5 DeepSeek-V4-Pro