Qwen3 VL 30B A3B Reasoning vs Qwen3 VL 32B Instruct
Wins 4 of 7 areas
Reasoning · Facts · Long documents · Following instructions
Wins 2 of 7 areas
Coding · Images and charts
Qwen3 VL 30B A3B Reasoning wins more areas, narrowly.Qwen3 VL 32B Instruct is cheaper and better at images and charts.
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.
- ReasoningHard problems that need careful thinking20Qwen3 VL 30B A3B Reasoning2 of 3 tests · 1 tie
- FactsGetting facts right instead of making them up20Qwen3 VL 30B A3B Reasoning2 of 2 tests
- Long documentsFinding answers in very long texts10Qwen3 VL 30B A3B Reasoning1 of 1 test
- Following instructionsDoing exactly what it is asked10Qwen3 VL 30B A3B Reasoning1 of 1 test
- Images and chartsUnderstanding pictures, charts and video06Qwen3 VL 32B Instruct6 of 6 tests
- CodingWriting and fixing software12Qwen3 VL 32B Instruct2 of 3 tests
- 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.
Qwen3 VL 32B Instruct costs 69% less for the same work.
The biggest differences
The three tests each model wins by the widest margin. Scores are out of 100.
Where Qwen3 VL 30B A3B Reasoning pulls ahead
- Recent programming contest problemsLiveCodeBench v6+20.4points ahead
- Reasons across sets of long documentsAA-LCR+8.4points ahead
- Follows unfamiliar, precisely checkable instructionsIFBench+5.9points ahead
Where Qwen3 VL 32B Instruct pulls ahead
- Reasoning about charts from research papersCharXiv (RQ)+6.2points ahead
- Reads text in imagesOCRBench+5.6points ahead
- Hard command-line tasks in a real terminalTerminal-Bench Hard+3points 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
- LiveCodeBench v6Qwen3 VL 30B A3B Reasoning by 20.464.243.8+20.4
- Terminal-Bench HardQwen3 VL 32B Instruct by 35.38.3+3
- SciCodeQwen3 VL 32B Instruct by 1.328.830.1+1.3
AgentsEven
- GDPValQwen3 VL 30B A3B Reasoning by 8.38.30+8.3
- OSWorld-VerifiedQwen3 VL 32B Instruct by 230.632.6+2
ReasoningQwen3 VL 30B A3B Reasoning
- GPQA DiamondQwen3 VL 30B A3B Reasoning by 4.97267.1+4.9
- Humanity's Last ExamQwen3 VL 30B A3B Reasoning by 2.18.96.8+2.1
- CritPttie00tie
FactsQwen3 VL 30B A3B Reasoning
- AA-Omniscience · AccuracyQwen3 VL 30B A3B Reasoning by 2.116.914.8+2.1
- AA-Omniscience · Non-hallucinationQwen3 VL 30B A3B Reasoning by 210.28.2+2
Images and chartsQwen3 VL 32B Instruct
Long documentsQwen3 VL 30B A3B Reasoning
- AA-LCRQwen3 VL 30B A3B Reasoning by 8.442.734.3+8.4
Following instructionsQwen3 VL 30B A3B Reasoning
- IFBenchQwen3 VL 30B A3B Reasoning by 5.945.139.2+5.9
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.
- AIME 2025Qwen3 VL 30B A3B Reasoning by 16.983.166.2+16.9
- τ²-Bench Telecom (AA run)Qwen3 VL 32B Instruct by 9.319.929.2+9.3
- Arena-Hard-V2Qwen3 VL 32B Instruct by 856.764.7+8
- AA-OmniscienceQwen3 VL 30B A3B Reasoning by 5.8-57.6-63.4+5.8
- AA Agentic IndexQwen3 VL 30B A3B Reasoning by 5.214.99.7+5.2
- OCRBench v2_enQwen3 VL 32B Instruct by 4.862.667.4+4.8
- LVBenchQwen3 VL 32B Instruct by 4.659.263.8+4.6
- ERQAQwen3 VL 32B Instruct by 3.545.348.8+3.5
- IFEvalQwen3 VL 32B Instruct by 381.784.7+3
- MMLU-ProXQwen3 VL 30B A3B Reasoning by 2.776.173.4+2.7
- AI2DQwen3 VL 32B Instruct by 2.686.989.5+2.6
- Artificial Analysis Coding IndexQwen3 VL 32B Instruct by 2.513.115.6+2.5
- CC-OCRQwen3 VL 32B Instruct by 2.577.880.3+2.5
- MATH-VisionQwen3 VL 30B A3B Reasoning by 2.365.763.4+2.3
- HallusionBenchQwen3 VL 30B A3B Reasoning by 2.26663.8+2.2
- DocVQA_testQwen3 VL 32B Instruct by 1.99596.9+1.9
- MMLU-ProQwen3 VL 30B A3B Reasoning by 1.980.578.6+1.9
- SuperGPQAQwen3 VL 30B A3B Reasoning by 1.856.454.6+1.8
- BFCL v3Qwen3 VL 32B Instruct by 1.668.670.2+1.6
- RealWorldQAQwen3 VL 32B Instruct by 1.677.479+1.6
- CharadesSTAQwen3 VL 30B A3B Reasoning by 1.562.761.2+1.5
- OCRBench v2 (Chinese)Qwen3 VL 30B A3B Reasoning by 1.260.459.2+1.2
- AA IntelligenceQwen3 VL 30B A3B Reasoning by 1.19.58.4+1.1
- MMLU-ReduxQwen3 VL 30B A3B Reasoning by 1.190.989.8+1.1
- ScreenSpotQwen3 VL 32B Instruct by 1.194.795.8+1.1
- InfoVQA_testQwen3 VL 32B Instruct by 18687+1
- MV-Benchtie7272.8tie
- ScreenSpot-Pro (No tools)tie57.357.9tie
- Includetie74.574tie
- MM MTBenchtie7.98.4tie
- MMMU (val) (Pass@1)tie7676tie
Questions people ask
Which is better, Qwen3 VL 30B A3B Reasoning or Qwen3 VL 32B Instruct?
Qwen3 VL 30B A3B Reasoning wins four of the seven areas where both have results: reasoning, facts, long documents and following instructions. Qwen3 VL 32B Instruct wins coding and images and charts, and costs 69% less. They are level on agents.
Which is better for coding?
Qwen3 VL 32B Instruct. It wins 2 of the 3 coding tests both models report; Qwen3 VL 30B A3B Reasoning wins 1.
Which is cheaper?
Qwen3 VL 30B A3B Reasoning costs $0.20 per million input tokens and $2.40 per million output tokens; Qwen3 VL 32B Instruct costs $0.16 and $0.64. That makes Qwen3 VL 32B Instruct about 69% 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.