Qwen3 VL 235B A22B Reasoning vs Qwen3 VL 32B Reasoning
Wins 4 of 7 areas
Coding · Reasoning · Images and charts · Long documents
Wins 2 of 7 areas
Agents · Following instructions
Qwen3 VL 235B A22B Reasoning wins more areas, narrowly.Qwen3 VL 32B Reasoning is cheaper and better at agents.
Scores updated · 45 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.
- CodingWriting and fixing software30Qwen3 VL 235B A22B Reasoning3 of 3 tests
- ReasoningHard problems that need careful thinking20Qwen3 VL 235B A22B Reasoning2 of 3 tests · 1 tie
- Images and chartsUnderstanding pictures, charts and video21Qwen3 VL 235B A22B Reasoning2 of 6 tests · 3 ties
- Long documentsFinding answers in very long texts10Qwen3 VL 235B A22B Reasoning1 of 1 test
- AgentsCarrying out multi-step tasks on its own01Qwen3 VL 32B Reasoning1 of 1 test
- Following instructionsDoing exactly what it is asked01Qwen3 VL 32B Reasoning1 of 1 test
- FactsGetting facts right instead of making them up11Even1 each
Agents, 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 Reasoning costs 82% 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 235B A22B Reasoning pulls ahead
Where Qwen3 VL 32B Reasoning pulls ahead
- Completes tasks by operating a computer desktopOSWorld-Verified+2.9points ahead
- Follows unfamiliar, precisely checkable instructionsIFBench+2.9points ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+1.5points ahead
Every test, side by side
All 45 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingQwen3 VL 235B A22B Reasoning
- SciCodeQwen3 VL 235B A22B Reasoning by 11.439.928.5+11.4
- LiveCodeBench v6Qwen3 VL 235B A22B Reasoning by 4.570.165.6+4.5
- Terminal-Bench HardQwen3 VL 235B A22B Reasoning by 3.811.47.6+3.8
AgentsQwen3 VL 32B Reasoning
- OSWorld-VerifiedQwen3 VL 32B Reasoning by 2.938.141+2.9
ReasoningQwen3 VL 235B A22B Reasoning
- GPQA DiamondQwen3 VL 235B A22B Reasoning by 3.977.273.3+3.9
- Humanity's Last ExamQwen3 VL 235B A22B Reasoning by 1.811.910.1+1.8
- CritPttie00tie
FactsEven
- AA-Omniscience · AccuracyQwen3 VL 235B A22B Reasoning by 3.920.917+3.9
- AA-Omniscience · Non-hallucinationQwen3 VL 32B Reasoning by 1.514.816.3+1.5
Images and chartsQwen3 VL 235B A22B Reasoning
Long documentsQwen3 VL 235B A22B Reasoning
- AA-LCRQwen3 VL 235B A22B Reasoning by 8.463.755.3+8.4
Following instructionsQwen3 VL 32B Reasoning
- IFBenchQwen3 VL 32B Reasoning by 2.956.559.4+2.9
Other results28 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.
- SimpleQAQwen3 VL 32B Reasoning by 1144.455.4+11
- τ²-Bench Telecom (AA run)Qwen3 VL 235B A22B Reasoning by 8.554.145.6+8.5
- AA-OmniscienceQwen3 VL 235B A22B Reasoning by 6-46.5-52.5+6
- AIME 2025Qwen3 VL 235B A22B Reasoning by 689.783.7+6
- SuperGPQAQwen3 VL 235B A22B Reasoning by 5.364.359+5.3
- ScreenSpot-Pro (No tools)Qwen3 VL 235B A22B Reasoning by 4.761.857.1+4.7
- MATH-VisionQwen3 VL 235B A22B Reasoning by 4.474.670.2+4.4
- IncludeQwen3 VL 235B A22B Reasoning by 3.78076.3+3.7
- MMLU-ProXQwen3 VL 235B A22B Reasoning by 3.480.677.2+3.4
- RealWorldQAQwen3 VL 235B A22B Reasoning by 2.981.378.4+2.9
- MMMU (val) (Pass@1)Qwen3 VL 235B A22B Reasoning by 2.580.678.1+2.5
- MMLU-ReduxQwen3 VL 235B A22B Reasoning by 1.893.791.9+1.8
- MMLU-ProQwen3 VL 235B A22B Reasoning by 1.783.882.1+1.7
- OCRBench v2_enQwen3 VL 32B Reasoning by 1.666.868.4+1.6
- AA IntelligenceQwen3 VL 235B A22B Reasoning by 1.513.411.9+1.5
- OCRBench v2 (Chinese)Qwen3 VL 235B A22B Reasoning by 1.463.562.1+1.4
- LVBenchQwen3 VL 235B A22B Reasoning by 163.662.6+1
- VideoMMMUQwen3 VL 235B A22B Reasoning by 18079+1
- CharadesSTAtie63.562.8tie
- HallusionBenchtie66.767.4tie
- DocVQA_testtie96.596.1tie
- IFEvaltie88.287.8tie
- AI2Dtie89.288.9tie
- InfoVQA_testtie89.589.2tie
- ScreenSpottie95.495.7tie
- BFCL v3tie71.971.7tie
- ERQAtie52.552.3tie
- MM MTBenchtie8.58.3tie
Questions people ask
Which is better, Qwen3 VL 235B A22B Reasoning or Qwen3 VL 32B Reasoning?
Qwen3 VL 235B A22B Reasoning wins four of the seven areas where both have results: coding, reasoning, images and charts and long documents. Qwen3 VL 32B Reasoning wins agents and following instructions, and costs 82% less. They are level on facts.
Which is better for coding?
Qwen3 VL 235B A22B Reasoning. It wins 3 of the 3 coding tests both models report; Qwen3 VL 32B Reasoning wins none.
Which is cheaper?
Qwen3 VL 235B A22B Reasoning costs $0.40 per million input tokens and $4.00 per million output tokens; Qwen3 VL 32B Reasoning costs $0.16 and $0.64. That makes Qwen3 VL 32B Reasoning about 82% cheaper for the same work.
How do you compare the two?
We use the 45 benchmark tests both models have published scores on. The verdict counts the 17 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 28 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.