Which is better, DeepSeek-V3 or Qwen3.5 2B?

DeepSeek-V3 vs Qwen3.5 2B

30 SHARED BENCHMARKS

Across 30 shared benchmarks, DeepSeek-V3 scores higher on 23 and Qwen3.5 2B on 5, with 2 level. The widest gap is Artificial Analysis Coding Index, where DeepSeek-V3 scores 23 against 2.9.

DEEPSEEKVSALIBABA30 SHARED23–5 HEAD-TO-HEAD

BenchmarkDeepSeek-V3MARGINQwen3.5 2B

AA Agentic Index1.60.8

AA Intelligence15.46.9

AA-LCR4125.6

AA-Omniscience-37.6-60.2

AIME 202459.417

AIME 202551.319

Artificial Analysis Coding Index232.9

C-Eval90.173.2

critpt00

gdpval00

GPQA Diamond68.451.6

HLE5.25

IFBench4141.3

ifeval87.378.6

longbench_v248.738.7

math_500_em9481.5

mmlu_prox70.552.3

mmlu_redux90.579.6

MMLU-Pro81.266.5

mmmlu79.463.1

multichallenge31.433.7

OmniScience Accuracy25.48.1

OmniScience Non-Hallucination23.325.8

scicode397.2

supergpqa53.737.5

TauBench V3 - Banking4.72.5

Terminal-Bench 2.116.93

Terminal-Bench Hard15.23.8

τ²-Bench Telecom (AA run)47.181.6

τ³-Bench4.76.2

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 DeepSeek-V3 withALL PAIRINGS →

DeepSeekDeepSeek-V4-Pro DeepSeek-V3.2 DeepSeek-V4-Flash

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 K2.6 Kimi K2.5 o3 GPT-5.1 Claude Opus 5 Gemma 4 26B A4B Kimi K3 Llama 3.1 Instruct 405B