Qwen3 VL 32B Reasoning vs Qwen3 VL 4B (Reasoning)
Wins 6 of 7 areas
Coding · Reasoning · Facts · Images and charts · Long documents · Following instructions
Wins 0 of 7 areas
—
Qwen3 VL 32B Reasoning is the stronger all-rounder.
Scores updated · 49 tests both models report · How we compare
Where each one wins
Tests won in each of the seven areas where both have results. Each piece is one test, so a longer bar means more evidence; grey means the two scored within a point of each other.
- Images and chartsUnderstanding pictures, charts and video60Qwen3 VL 32B Reasoning6 of 6 tests
- CodingWriting and fixing software30Qwen3 VL 32B Reasoning3 of 3 tests
- ReasoningHard problems that need careful thinking20Qwen3 VL 32B Reasoning2 of 3 tests · 1 tie
- FactsGetting facts right instead of making them up20Qwen3 VL 32B Reasoning2 of 2 tests
- Long documentsFinding answers in very long texts10Qwen3 VL 32B Reasoning1 of 1 test
- Following instructionsDoing exactly what it is asked10Qwen3 VL 32B Reasoning1 of 1 test
- AgentsCarrying out multi-step tasks on its own11Even1 each
Long documents and following instructions rest on a single test each.
What it costs
Prices per million tokens, roughly 750,000 words. The bars show the cost of a million tokens read plus a million written.
The biggest differences
The tests each model wins by the widest margin, up to three each. Scores are out of 100.
Where Qwen3 VL 32B Reasoning pulls ahead
- Reasons across sets of long documentsAA-LCR+34points ahead
- Graduate-level biology, physics and chemistry questionsGPQA Diamond+23.9points ahead
- Follows unfamiliar, precisely checkable instructionsIFBench+22.8points ahead
Where Qwen3 VL 4B (Reasoning) pulls ahead
- Real work tasks from 44 professionsGDPVal+6.5points ahead
Every test, side by side
All 49 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingQwen3 VL 32B Reasoning
- LiveCodeBench v6Qwen3 VL 32B Reasoning by 14.365.651.3+14.3
- SciCodeQwen3 VL 32B Reasoning by 11.428.517.1+11.4
- Terminal-Bench HardQwen3 VL 32B Reasoning by 6.17.61.5+6.1
AgentsEven
- OSWorld-VerifiedQwen3 VL 32B Reasoning by 9.64131.4+9.6
- GDPValQwen3 VL 4B (Reasoning) by 6.57.313.8+6.5
ReasoningQwen3 VL 32B Reasoning
- GPQA DiamondQwen3 VL 32B Reasoning by 23.973.349.4+23.9
- Humanity's Last ExamQwen3 VL 32B Reasoning by 5.510.14.6+5.5
- CritPttie00tie
FactsQwen3 VL 32B Reasoning
- AA-Omniscience · Non-hallucinationQwen3 VL 32B Reasoning by 8.116.38.2+8.1
- AA-Omniscience · AccuracyQwen3 VL 32B Reasoning by 4.81712.2+4.8
Images and chartsQwen3 VL 32B Reasoning
Long documentsQwen3 VL 32B Reasoning
- AA-LCRQwen3 VL 32B Reasoning by 3455.321.3+34
Following instructionsQwen3 VL 32B Reasoning
- IFBenchQwen3 VL 32B Reasoning by 22.859.436.6+22.8
Other results31 tests, not counted
Tests outside the eight areas. They are not counted above: several are summary scores built from other tests, or the same test under another name.
- τ²-Bench Telecom (AA run)Qwen3 VL 32B Reasoning by 30.145.615.5+30.1
- Arena-Hard-V2Qwen3 VL 32B Reasoning by 23.760.536.8+23.7
- AA-OmniscienceQwen3 VL 32B Reasoning by 15.9-52.5-68.4+15.9
- MMLU-ProXQwen3 VL 32B Reasoning by 12.277.265+12.2
- SuperGPQAQwen3 VL 32B Reasoning by 12.25946.8+12.2
- IncludeQwen3 VL 32B Reasoning by 11.776.364.6+11.7
- MATH-VisionQwen3 VL 32B Reasoning by 10.270.260+10.2
- VideoMMMUQwen3 VL 32B Reasoning by 9.67969.4+9.6
- AIME 2025Qwen3 VL 32B Reasoning by 9.283.774.5+9.2
- LVBenchQwen3 VL 32B Reasoning by 9.162.653.5+9.1
- AA Agentic IndexQwen3 VL 32B Reasoning by 923.414.4+9
- MMLU-ProQwen3 VL 32B Reasoning by 8.582.173.6+8.5
- ScreenSpot-Pro (No tools)Qwen3 VL 32B Reasoning by 7.957.149.2+7.9
- Artificial Analysis Coding IndexQwen3 VL 32B Reasoning by 7.814.56.7+7.8
- MMMU (val) (Pass@1)Qwen3 VL 32B Reasoning by 7.378.170.8+7.3
- OCRBench v2_enQwen3 VL 32B Reasoning by 6.668.461.8+6.6
- OCRBench v2 (Chinese)Qwen3 VL 32B Reasoning by 6.362.155.8+6.3
- InfoVQA_testQwen3 VL 32B Reasoning by 6.289.283+6.2
- MMLU-ReduxQwen3 VL 32B Reasoning by 5.991.986+5.9
- IFEvalQwen3 VL 32B Reasoning by 5.287.882.6+5.2
- RealWorldQAQwen3 VL 32B Reasoning by 5.278.473.2+5.2
- ERQAQwen3 VL 32B Reasoning by 552.347.3+5
- AA IntelligenceQwen3 VL 32B Reasoning by 4.911.97+4.9
- BFCL v3Qwen3 VL 32B Reasoning by 4.471.767.3+4.4
- AI2DQwen3 VL 32B Reasoning by 488.984.9+4
- MV-BenchQwen3 VL 32B Reasoning by 3.973.269.3+3.9
- CharadesSTAQwen3 VL 32B Reasoning by 3.862.859+3.8
- HallusionBenchQwen3 VL 32B Reasoning by 3.367.464.1+3.3
- ScreenSpotQwen3 VL 32B Reasoning by 2.895.792.9+2.8
- DocVQA_testQwen3 VL 32B Reasoning by 1.996.194.2+1.9
- MM MTBenchtie8.37.7tie
Questions people ask
Which is better, Qwen3 VL 32B Reasoning or Qwen3 VL 4B (Reasoning)?
Qwen3 VL 32B Reasoning wins six of the seven areas where both have results: coding, reasoning, facts, images and charts, long documents and following instructions. Qwen3 VL 4B (Reasoning) wins none. They are level on agents.
Which is better for coding?
Qwen3 VL 32B Reasoning. It wins 3 of the 3 coding tests both models report; Qwen3 VL 4B (Reasoning) wins none.
How do you compare the two?
We use the 49 benchmark tests both models have published scores on. The verdict counts the 18 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 31 are listed but not counted, because several are summary scores or repeat a test. Each score is the one shown on the model's own page.