Qwen3 VL 4B (Reasoning) vs Qwen3 VL Thinking (8B)
Wins 0 of 7 areas
—
Wins 6 of 7 areas
Coding · Reasoning · Facts · Images and charts · Long documents · Following instructions
Qwen3 VL Thinking (8B) is the stronger all-rounder.
Scores updated · 55 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 video07Qwen3 VL Thinking (8B)7 of 7 tests
- CodingWriting and fixing software03Qwen3 VL Thinking (8B)3 of 3 tests
- ReasoningHard problems that need careful thinking01Qwen3 VL Thinking (8B)1 of 3 tests · 2 ties
- FactsGetting facts right instead of making them up01Qwen3 VL Thinking (8B)1 of 2 tests · 1 tie
- Long documentsFinding answers in very long texts01Qwen3 VL Thinking (8B)1 of 1 test
- Following instructionsDoing exactly what it is asked01Qwen3 VL Thinking (8B)1 of 1 test
- 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.
The biggest differences
The tests each model wins by the widest margin, up to three each. Scores are out of 100.
Where Qwen3 VL Thinking (8B) pulls ahead
- Questions about short and long videosVideo-MME+12.1points ahead
- Reasons across sets of long documentsAA-LCR+12points ahead
- Graduate-level biology, physics and chemistry questionsGPQA Diamond+8.5points ahead
Where Qwen3 VL 4B (Reasoning) pulls ahead
- Real work tasks from 44 professionsGDPVal+5.4points ahead
Every test, side by side
All 55 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingQwen3 VL Thinking (8B)
- LiveCodeBench v6Qwen3 VL Thinking (8B) by 7.351.358.6+7.3
- SciCodeQwen3 VL Thinking (8B) by 4.817.121.9+4.8
- Terminal-Bench HardQwen3 VL Thinking (8B) by 2.31.53.8+2.3
AgentsEven
- GDPValQwen3 VL 4B (Reasoning) by 5.413.88.4+5.4
- OSWorld-VerifiedQwen3 VL Thinking (8B) by 2.531.433.9+2.5
ReasoningQwen3 VL Thinking (8B)
- GPQA DiamondQwen3 VL Thinking (8B) by 8.549.457.9+8.5
- Humanity's Last Examtie4.63.8tie
- CritPttie00.3tie
FactsQwen3 VL Thinking (8B)
- AA-Omniscience · AccuracyQwen3 VL Thinking (8B) by 8.212.220.4+8.2
- AA-Omniscience · Non-hallucinationtie8.28.6tie
Images and chartsQwen3 VL Thinking (8B)
- Video-MMEQwen3 VL Thinking (8B) by 12.159.771.8+12.1
- BLINKQwen3 VL Thinking (8B) by 5.363.468.7+5.3
- MMMU-ProQwen3 VL Thinking (8B) by 4.65256.6+4.6
- CharXiv (RQ)Qwen3 VL Thinking (8B) by 2.750.353+2.7
- MMStarQwen3 VL Thinking (8B) by 2.173.275.3+2.1
- MathVistaQwen3 VL Thinking (8B) by 1.979.581.4+1.9
- OCRBenchQwen3 VL Thinking (8B) by 1.180.881.9+1.1
Long documentsQwen3 VL Thinking (8B)
- AA-LCRQwen3 VL Thinking (8B) by 1221.333.3+12
Following instructionsQwen3 VL Thinking (8B)
- IFBenchQwen3 VL Thinking (8B) by 3.336.639.9+3.3
Other results36 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.
- AA-OmniscienceQwen3 VL Thinking (8B) by 16.1-68.4-52.3+16.1
- Arena-Hard-V2Qwen3 VL Thinking (8B) by 14.336.851.1+14.3
- AIME 2024Qwen3 VL Thinking (8B) by 13.172.986+13.1
- HMMT 2025Qwen3 VL Thinking (8B) by 7.553.160.6+7.5
- τ²-Bench Telecom (AA run)Qwen3 VL Thinking (8B) by 715.522.5+7
- LiveCodeBenchQwen3 VL Thinking (8B) by 6.751.358+6.7
- AIME 2025Qwen3 VL Thinking (8B) by 5.874.580.3+5.8
- MMLU-ProXQwen3 VL Thinking (8B) by 5.76570.7+5.7
- IncludeQwen3 VL Thinking (8B) by 4.964.669.5+4.9
- SuperGPQAQwen3 VL Thinking (8B) by 4.446.851.2+4.4
- BFCL v3Qwen3 VL 4B (Reasoning) by 4.367.363+4.3
- MMLU-ProQwen3 VL Thinking (8B) by 3.773.677.3+3.7
- OCRBench v2 (Chinese)Qwen3 VL Thinking (8B) by 3.455.859.2+3.4
- VideoMMMUQwen3 VL Thinking (8B) by 3.469.472.8+3.4
- MMMU (val) (Pass@1)Qwen3 VL Thinking (8B) by 3.370.874.1+3.3
- Artificial Analysis Coding IndexQwen3 VL Thinking (8B) by 3.16.79.8+3.1
- InfoVQA_testQwen3 VL Thinking (8B) by 38386+3
- MMLU-ReduxQwen3 VL Thinking (8B) by 2.88688.8+2.8
- MATH-VisionQwen3 VL Thinking (8B) by 2.76062.7+2.7
- ScreenSpot-Pro (No tools)Qwen3 VL 4B (Reasoning) by 2.649.246.6+2.6
- CC-OCRQwen3 VL Thinking (8B) by 2.573.876.3+2.5
- LVBenchQwen3 VL Thinking (8B) by 2.353.555.8+2.3
- OCRBench v2_enQwen3 VL Thinking (8B) by 2.161.863.9+2.1
- HallusionBenchQwen3 VL Thinking (8B) by 1.364.165.4+1.3
- AA Agentic IndexQwen3 VL Thinking (8B) by 1.214.415.6+1.2
- AA IntelligenceQwen3 VL Thinking (8B) by 1.278.2+1.2
- DocVQA_testQwen3 VL Thinking (8B) by 1.194.295.3+1.1
- CharadesSTAtie5959.9tie
- All-Angles-Benchtie46.745.9tie
- ScreenSpottie92.993.6tie
- IFEvaltie82.683.2tie
- ERQAtie47.346.8tie
- MM MTBenchtie7.78tie
- MV-Benchtie69.369tie
- RealWorldQAtie73.273.5tie
- AI2Dtie84.984.9tie
Questions people ask
Which is better, Qwen3 VL 4B (Reasoning) or Qwen3 VL Thinking (8B)?
Qwen3 VL Thinking (8B) wins six of the seven areas where both have results: coding, reasoning, facts, images and charts, long documents and following instructions. Qwen3 VL 4B (Reasoning) wins none. They are level on agents.
Which is better for coding?
Qwen3 VL Thinking (8B). It wins 3 of the 3 coding tests both models report; Qwen3 VL 4B (Reasoning) wins none.
How do you compare the two?
We use the 55 benchmark tests both models have published scores on. The verdict counts the 19 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 36 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.