Qwen3 VL 4B Instruct vs Qwen3 VL 8B Instruct
Wins 0 of 7 areas
—
Wins 6 of 7 areas
Coding · Agents · Reasoning · Facts · Images and charts · Long documents
Qwen3 VL 8B Instruct is the stronger all-rounder.
Scores updated · 52 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 video06Qwen3 VL 8B Instruct6 of 6 tests
- CodingWriting and fixing software03Qwen3 VL 8B Instruct3 of 3 tests
- AgentsCarrying out multi-step tasks on its own02Qwen3 VL 8B Instruct2 of 2 tests
- FactsGetting facts right instead of making them up02Qwen3 VL 8B Instruct2 of 2 tests
- ReasoningHard problems that need careful thinking01Qwen3 VL 8B Instruct1 of 3 tests · 2 ties
- Long documentsFinding answers in very long texts01Qwen3 VL 8B Instruct1 of 1 test
- Following instructionsDoing exactly what it is asked00Even0 each · 1 tie
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 8B Instruct pulls ahead
- Answers hard knowledge questions correctlyAA-Omniscience · Accuracy+9.6points ahead
- Completes tasks by operating a computer desktopOSWorld-Verified+7.7points ahead
- Reasoning about charts from research papersCharXiv (RQ)+6.7points ahead
Where Qwen3 VL 4B Instruct pulls ahead
No clear win on a test scored out of 100.
Every test, side by side
All 52 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingQwen3 VL 8B Instruct
- SciCodeQwen3 VL 8B Instruct by 3.713.717.4+3.7
- Terminal-Bench HardQwen3 VL 8B Instruct by 2.302.3+2.3
- LiveCodeBench v6Qwen3 VL 8B Instruct by 1.437.939.3+1.4
AgentsQwen3 VL 8B Instruct
- GDPValQwen3 VL 8B Instruct by 9.109.1+9.1
- OSWorld-VerifiedQwen3 VL 8B Instruct by 7.726.233.9+7.7
ReasoningQwen3 VL 8B Instruct
- GPQA DiamondQwen3 VL 8B Instruct by 5.637.142.7+5.6
- Humanity's Last Examtie3.62.7tie
- CritPttie00tie
FactsQwen3 VL 8B Instruct
- AA-Omniscience · AccuracyQwen3 VL 8B Instruct by 9.610.920.5+9.6
- AA-Omniscience · Non-hallucinationQwen3 VL 8B Instruct by 6.42.69+6.4
Images and chartsQwen3 VL 8B Instruct
Long documentsQwen3 VL 8B Instruct
- AA-LCRQwen3 VL 8B Instruct by 2.71416.7+2.7
Other results34 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 8B Instruct by 24.1-76-51.9+24.1
- AA Agentic IndexQwen3 VL 8B Instruct by 10.67.818.4+10.6
- WildBenchQwen3 VL 8B Instruct by 9.556.866.3+9.5
- VideoMMMUQwen3 VL 8B Instruct by 9.156.265.3+9.1
- Arena HardQwen3 VL 8B Instruct by 943.852.8+9
- MMLU-ProXQwen3 VL 8B Instruct by 659.465.4+6
- τ²-Bench Telecom (AA run)Qwen3 VL 8B Instruct by 5.823.429.2+5.8
- IncludeQwen3 VL 8B Instruct by 5.661.467+5.6
- ScreenSpot-Pro (No tools)Qwen3 VL 4B Instruct by 4.959.554.6+4.9
- ERQAQwen3 VL 8B Instruct by 4.541.345.8+4.5
- MMLU-ProQwen3 VL 8B Instruct by 4.567.171.6+4.5
- SuperGPQAQwen3 VL 8B Instruct by 4.240.344.5+4.2
- CC-OCRQwen3 VL 8B Instruct by 3.776.279.9+3.7
- OCRBench v2 (Chinese)Qwen3 VL 8B Instruct by 3.657.661.2+3.6
- HallusionBenchQwen3 VL 8B Instruct by 3.557.661.1+3.5
- MMLU-ReduxQwen3 VL 8B Instruct by 3.481.584.9+3.4
- BFCL v3Qwen3 VL 8B Instruct by 363.366.3+3
- InfoVQA_testQwen3 VL 8B Instruct by 2.880.383.1+2.8
- Artificial Analysis Coding IndexQwen3 VL 8B Instruct by 2.74.67.3+2.7
- MATH-VisionQwen3 VL 8B Instruct by 2.351.653.9+2.3
- MMMU (val) (Pass@1)Qwen3 VL 8B Instruct by 2.267.469.6+2.2
- HMMT 2025Qwen3 VL 8B Instruct by 1.830.732.5+1.8
- LVBenchQwen3 VL 8B Instruct by 1.856.258+1.8
- OCRBench v2_enQwen3 VL 8B Instruct by 1.763.765.4+1.7
- AA IntelligenceQwen3 VL 8B Instruct by 1.65.77.3+1.6
- AI2DQwen3 VL 8B Instruct by 1.684.185.7+1.6
- IFEvalQwen3 VL 8B Instruct by 1.482.383.7+1.4
- DocVQA_testtie95.396.1tie
- AIME 2025tie46.645.9tie
- RealWorldQAtie70.971.5tie
- CharadesSTAtie55.556tie
- ScreenSpottie9494.4tie
- MM MTBenchtie7.57.7tie
- MV-Benchtie68.968.7tie
Questions people ask
Which is better, Qwen3 VL 4B Instruct or Qwen3 VL 8B Instruct?
Qwen3 VL 8B Instruct wins six of the seven areas where both have results: coding, agents, reasoning, facts, images and charts and long documents. Qwen3 VL 4B Instruct wins none. They are level on following instructions.
Which is better for coding?
Qwen3 VL 8B Instruct. It wins 3 of the 3 coding tests both models report; Qwen3 VL 4B Instruct wins none.
How do you compare the two?
We use the 52 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 34 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.