Qwen3 VL 30B A3B Reasoning vs Step3 VL 10B
Wins 2 of 7 areas
Agents · Long documents
Wins 3 of 7 areas
Coding · Images and charts · Following instructions
Step3 VL 10B wins more areas, narrowly.Qwen3 VL 30B A3B Reasoning is better at agents.
Scores updated · 25 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 video05Step3 VL 10B5 of 5 tests
- CodingWriting and fixing software01Step3 VL 10B1 of 2 tests · 1 tie
- Following instructionsDoing exactly what it is asked01Step3 VL 10B1 of 1 test
- AgentsCarrying out multi-step tasks on its own10Qwen3 VL 30B A3B Reasoning1 of 1 test
- Long documentsFinding answers in very long texts10Qwen3 VL 30B A3B Reasoning1 of 1 test
- ReasoningHard problems that need careful thinking11Even1 each · 1 tie
- 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.
The biggest differences
The tests each model wins by the widest margin, up to three each. Scores are out of 100.
Where Step3 VL 10B pulls ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+6.5points ahead
- Reads text in imagesOCRBench+5.1points ahead
- Follows unfamiliar, precisely checkable instructionsIFBench+5.1points ahead
Where Qwen3 VL 30B A3B Reasoning pulls ahead
- Answers hard knowledge questions correctlyAA-Omniscience · Accuracy+3.6points ahead
- Graduate-level biology, physics and chemistry questionsGPQA Diamond+3points ahead
Every test, side by side
All 25 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingStep3 VL 10B
- SciCodeStep3 VL 10B by 2.328.831.1+2.3
- Terminal-Bench Hardtie5.35.3tie
ReasoningEven
- GPQA DiamondQwen3 VL 30B A3B Reasoning by 37269+3
- Humanity's Last ExamStep3 VL 10B by 1.98.910.8+1.9
- CritPttie00tie
FactsEven
- AA-Omniscience · Non-hallucinationStep3 VL 10B by 6.510.216.7+6.5
- AA-Omniscience · AccuracyQwen3 VL 30B A3B Reasoning by 3.616.913.3+3.6
Images and chartsStep3 VL 10B
Long documentsQwen3 VL 30B A3B Reasoning
- AA-LCRQwen3 VL 30B A3B Reasoning by 42.742.70+42.7
Following instructionsStep3 VL 10B
- IFBenchStep3 VL 10B by 5.145.150.2+5.1
Other results10 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.
- HMMT 2025Step3 VL 10B by 24.567.692.1+24.5
- AA Agentic IndexQwen3 VL 30B A3B Reasoning by 9.514.95.4+9.5
- ScreenSpot-Pro (No tools)Qwen3 VL 30B A3B Reasoning by 5.757.351.5+5.7
- MATH-VisionStep3 VL 10B by 5.165.770.8+5.1
- AIME 2025Step3 VL 10B by 4.683.187.7+4.6
- τ²-Bench Telecom (AA run)Qwen3 VL 30B A3B Reasoning by 3.819.916.1+3.8
- AI2DStep3 VL 10B by 2.586.989.3+2.5
- AA IntelligenceQwen3 VL 30B A3B Reasoning by 1.89.57.7+1.8
- AA-OmniscienceQwen3 VL 30B A3B Reasoning by 1.4-57.6-59+1.4
- Artificial Analysis Coding Indextie13.113.9tie
Questions people ask
Which is better, Qwen3 VL 30B A3B Reasoning or Step3 VL 10B?
Step3 VL 10B wins three of the seven areas where both have results: coding, images and charts and following instructions. Qwen3 VL 30B A3B Reasoning wins agents and long documents. They are level on reasoning and facts.
Which is better for coding?
Step3 VL 10B. It wins 1 of the 2 coding tests both models report; Qwen3 VL 30B A3B Reasoning wins none, and 1 is a tie.
How do you compare the two?
We use the 25 benchmark tests both models have published scores on. The verdict counts the 15 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 10 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.