Qwen3 VL 30B A3B Instruct vs Qwen3 VL 32B Instruct
Wins 1 of 7 areas
Reasoning
Wins 6 of 7 areas
Coding · Agents · Facts · Images and charts · Long documents · Following instructions
Qwen3 VL 32B Instruct is the stronger all-rounder.Qwen3 VL 30B A3B Instruct is better at reasoning.
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 video04Qwen3 VL 32B Instruct4 of 6 tests · 2 ties
- CodingWriting and fixing software02Qwen3 VL 32B Instruct2 of 3 tests · 1 tie
- AgentsCarrying out multi-step tasks on its own01Qwen3 VL 32B Instruct1 of 2 tests · 1 tie
- FactsGetting facts right instead of making them up01Qwen3 VL 32B Instruct1 of 2 tests · 1 tie
- Long documentsFinding answers in very long texts01Qwen3 VL 32B Instruct1 of 1 test
- Following instructionsDoing exactly what it is asked01Qwen3 VL 32B Instruct1 of 1 test
- ReasoningHard problems that need careful thinking10Qwen3 VL 30B A3B Instruct1 of 3 tests · 2 ties
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.
Qwen3 VL 32B Instruct costs 20% less for the same work.
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
- Reasoning about charts from research papersCharXiv (RQ)+13.9points ahead
- Reasons across sets of long documentsAA-LCR+7points ahead
- Follows unfamiliar, precisely checkable instructionsIFBench+6.1points ahead
Where Qwen3 VL 30B A3B Instruct pulls ahead
- Graduate-level biology, physics and chemistry questionsGPQA Diamond+2.4points 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 Instruct
- Terminal-Bench HardQwen3 VL 32B Instruct by 2.26.18.3+2.2
- LiveCodeBench v6Qwen3 VL 32B Instruct by 1.242.643.8+1.2
- SciCodetie30.830.1tie
AgentsQwen3 VL 32B Instruct
- OSWorld-VerifiedQwen3 VL 32B Instruct by 2.330.332.6+2.3
- GDPValtie00tie
ReasoningQwen3 VL 30B A3B Instruct
- GPQA DiamondQwen3 VL 30B A3B Instruct by 2.469.567.1+2.4
- Humanity's Last Examtie6.36.8tie
- CritPttie00tie
FactsQwen3 VL 32B Instruct
- AA-Omniscience · Non-hallucinationQwen3 VL 32B Instruct by 1.96.38.2+1.9
- AA-Omniscience · Accuracytie15.614.8tie
Images and chartsQwen3 VL 32B Instruct
Long documentsQwen3 VL 32B Instruct
- AA-LCRQwen3 VL 32B Instruct by 727.334.3+7
Following instructionsQwen3 VL 32B Instruct
- IFBenchQwen3 VL 32B Instruct by 6.133.139.2+6.1
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 Instruct by 10.21929.2+10.2
- Arena-Hard-V2Qwen3 VL 32B Instruct by 6.258.564.7+6.2
- ERQAQwen3 VL 32B Instruct by 5.84348.8+5.8
- RealWorldQAQwen3 VL 32B Instruct by 5.373.779+5.3
- InfoVQA_testQwen3 VL 32B Instruct by 58287+5
- AI2DQwen3 VL 32B Instruct by 4.58589.5+4.5
- OCRBench v2_enQwen3 VL 32B Instruct by 4.263.267.4+4.2
- BFCL v3Qwen3 VL 32B Instruct by 3.966.370.2+3.9
- MATH-VisionQwen3 VL 32B Instruct by 3.260.263.4+3.2
- AIME 2025Qwen3 VL 30B A3B Instruct by 3.169.366.2+3.1
- ScreenSpot-Pro (No tools)Qwen3 VL 30B A3B Instruct by 2.660.557.9+2.6
- MMLU-ProXQwen3 VL 32B Instruct by 2.570.973.4+2.5
- IncludeQwen3 VL 32B Instruct by 2.471.674+2.4
- CharadesSTAQwen3 VL 30B A3B Instruct by 2.363.561.2+2.3
- HallusionBenchQwen3 VL 32B Instruct by 2.361.563.8+2.3
- DocVQA_testQwen3 VL 32B Instruct by 1.99596.9+1.9
- MMMU (val) (Pass@1)Qwen3 VL 32B Instruct by 1.874.276+1.8
- SuperGPQAQwen3 VL 32B Instruct by 1.553.154.6+1.5
- MMLU-ReduxQwen3 VL 32B Instruct by 1.488.489.8+1.4
- OCRBench v2 (Chinese)Qwen3 VL 32B Instruct by 1.457.859.2+1.4
- Artificial Analysis Coding IndexQwen3 VL 32B Instruct by 1.314.315.6+1.3
- LVBenchQwen3 VL 32B Instruct by 1.362.563.8+1.3
- IFEvalQwen3 VL 30B A3B Instruct by 1.185.884.7+1.1
- ScreenSpotQwen3 VL 32B Instruct by 1.194.795.8+1.1
- MMLU-Protie77.878.6tie
- AA Intelligencetie7.98.4tie
- MV-Benchtie72.372.8tie
- CC-OCRtie80.780.3tie
- MM MTBenchtie8.18.4tie
- AA Agentic Indextie9.59.7tie
- AA-Omnisciencetie-63.4-63.4tie
Questions people ask
Which is better, Qwen3 VL 30B A3B Instruct or Qwen3 VL 32B Instruct?
Qwen3 VL 32B Instruct wins six of the seven areas where both have results: coding, agents, facts, images and charts, long documents and following instructions. Qwen3 VL 30B A3B Instruct wins reasoning.
Which is better for coding?
Qwen3 VL 32B Instruct. It wins 2 of the 3 coding tests both models report; Qwen3 VL 30B A3B Instruct wins none, and 1 is a tie.
Which is cheaper?
Qwen3 VL 30B A3B Instruct costs $0.20 per million input tokens and $0.80 per million output tokens; Qwen3 VL 32B Instruct costs $0.16 and $0.64. That makes Qwen3 VL 32B Instruct about 20% cheaper for the same work.
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.