Qwen3 VL 32B Instruct vs Qwen3 VL 4B Instruct
Wins 7 of 7 areas
Coding · Agents · Reasoning · Facts · Images and charts · Long documents · Following instructions
Wins 0 of 7 areas
—
Qwen3 VL 32B Instruct is the stronger all-rounder.
Scores updated · 48 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 Instruct6 of 6 tests
- CodingWriting and fixing software30Qwen3 VL 32B Instruct3 of 3 tests
- ReasoningHard problems that need careful thinking20Qwen3 VL 32B Instruct2 of 3 tests · 1 tie
- FactsGetting facts right instead of making them up20Qwen3 VL 32B Instruct2 of 2 tests
- AgentsCarrying out multi-step tasks on its own10Qwen3 VL 32B Instruct1 of 2 tests · 1 tie
- Long documentsFinding answers in very long texts10Qwen3 VL 32B Instruct1 of 1 test
- Following instructionsDoing exactly what it is asked10Qwen3 VL 32B Instruct1 of 1 test
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 Instruct pulls ahead
- Graduate-level biology, physics and chemistry questionsGPQA Diamond+30points ahead
- Reasoning about charts from research papersCharXiv (RQ)+23.1points ahead
- Harder college exam questions with imagesMMMU-Pro+20.4points ahead
Where Qwen3 VL 4B Instruct pulls ahead
No clear win on a test scored out of 100.
Every test, side by side
All 48 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingQwen3 VL 32B Instruct
- SciCodeQwen3 VL 32B Instruct by 16.430.113.7+16.4
- Terminal-Bench HardQwen3 VL 32B Instruct by 8.38.30+8.3
- LiveCodeBench v6Qwen3 VL 32B Instruct by 5.943.837.9+5.9
AgentsQwen3 VL 32B Instruct
- OSWorld-VerifiedQwen3 VL 32B Instruct by 6.432.626.2+6.4
- GDPValtie00tie
ReasoningQwen3 VL 32B Instruct
- GPQA DiamondQwen3 VL 32B Instruct by 3067.137.1+30
- Humanity's Last ExamQwen3 VL 32B Instruct by 3.26.83.6+3.2
- CritPttie00tie
FactsQwen3 VL 32B Instruct
- AA-Omniscience · Non-hallucinationQwen3 VL 32B Instruct by 5.68.22.6+5.6
- AA-Omniscience · AccuracyQwen3 VL 32B Instruct by 3.914.810.9+3.9
Images and chartsQwen3 VL 32B Instruct
Long documentsQwen3 VL 32B Instruct
- AA-LCRQwen3 VL 32B Instruct by 20.334.314+20.3
Following instructionsQwen3 VL 32B Instruct
- IFBenchQwen3 VL 32B Instruct by 7.439.231.8+7.4
Other results30 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.
- AIME 2025Qwen3 VL 32B Instruct by 19.666.246.6+19.6
- SuperGPQAQwen3 VL 32B Instruct by 14.354.640.3+14.3
- MMLU-ProXQwen3 VL 32B Instruct by 1473.459.4+14
- AA-OmniscienceQwen3 VL 32B Instruct by 12.6-63.4-76+12.6
- IncludeQwen3 VL 32B Instruct by 12.67461.4+12.6
- MATH-VisionQwen3 VL 32B Instruct by 11.863.451.6+11.8
- MMLU-ProQwen3 VL 32B Instruct by 11.578.667.1+11.5
- Artificial Analysis Coding IndexQwen3 VL 32B Instruct by 1115.64.6+11
- MMMU (val) (Pass@1)Qwen3 VL 32B Instruct by 8.67667.4+8.6
- MMLU-ReduxQwen3 VL 32B Instruct by 8.389.881.5+8.3
- RealWorldQAQwen3 VL 32B Instruct by 8.17970.9+8.1
- LVBenchQwen3 VL 32B Instruct by 7.663.856.2+7.6
- ERQAQwen3 VL 32B Instruct by 7.548.841.3+7.5
- BFCL v3Qwen3 VL 32B Instruct by 6.970.263.3+6.9
- InfoVQA_testQwen3 VL 32B Instruct by 6.78780.3+6.7
- HallusionBenchQwen3 VL 32B Instruct by 6.263.857.6+6.2
- τ²-Bench Telecom (AA run)Qwen3 VL 32B Instruct by 5.829.223.4+5.8
- CharadesSTAQwen3 VL 32B Instruct by 5.761.255.5+5.7
- AI2DQwen3 VL 32B Instruct by 5.489.584.1+5.4
- CC-OCRQwen3 VL 32B Instruct by 4.180.376.2+4.1
- MV-BenchQwen3 VL 32B Instruct by 3.972.868.9+3.9
- OCRBench v2_enQwen3 VL 32B Instruct by 3.767.463.7+3.7
- AA IntelligenceQwen3 VL 32B Instruct by 2.78.45.7+2.7
- IFEvalQwen3 VL 32B Instruct by 2.484.782.3+2.4
- AA Agentic IndexQwen3 VL 32B Instruct by 1.99.77.8+1.9
- ScreenSpotQwen3 VL 32B Instruct by 1.895.894+1.8
- DocVQA_testQwen3 VL 32B Instruct by 1.696.995.3+1.6
- OCRBench v2 (Chinese)Qwen3 VL 32B Instruct by 1.659.257.6+1.6
- ScreenSpot-Pro (No tools)Qwen3 VL 4B Instruct by 1.657.959.5+1.6
- MM MTBenchtie8.47.5tie
Questions people ask
Which is better, Qwen3 VL 32B Instruct or Qwen3 VL 4B Instruct?
Qwen3 VL 32B Instruct wins all seven areas where both have results: coding, agents, reasoning, facts, images and charts, long documents and following instructions. Qwen3 VL 4B Instruct wins none.
Which is better for coding?
Qwen3 VL 32B Instruct. It wins 3 of the 3 coding tests both models report; Qwen3 VL 4B Instruct wins none.
How do you compare the two?
We use the 48 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 30 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.