Qwen3.5 27B vs Qwen3.5 4B
Wins 8 of 8 areas
Coding · Agents · Reasoning · Facts · Images and charts · Math · Long documents · Following instructions
Wins 0 of 8 areas
—
Qwen3.5 27B is the stronger all-rounder.Qwen3.5 4B is cheaper.
Scores updated · 45 tests both models report · How we compare
Where each one wins
Tests won in each of the eight areas we test. 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 video50Qwen3.5 27B5 of 5 tests
- CodingWriting and fixing software40Qwen3.5 27B4 of 4 tests
- ReasoningHard problems that need careful thinking20Qwen3.5 27B2 of 3 tests · 1 tie
- FactsGetting facts right instead of making them up20Qwen3.5 27B2 of 2 tests
- MathCompetition and research-level math20Qwen3.5 27B2 of 2 tests
- Long documentsFinding answers in very long texts20Qwen3.5 27B2 of 2 tests
- Following instructionsDoing exactly what it is asked20Qwen3.5 27B2 of 2 tests
- AgentsCarrying out multi-step tasks on its own10Qwen3.5 27B1 of 1 test
Agents rests on a single test.
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.
Qwen3.5 4B costs 93% less for the same work.
The biggest differences
The tests each model wins by the widest margin, up to three each. Scores are out of 100.
Where Qwen3.5 27B pulls ahead
- Multi-step tasks using many real software toolsMCP Atlas+27points ahead
- Recent programming contest problemsLiveCodeBench v6+24.9points ahead
- Math problems shown in pictures and chartsMathVista+24.2points ahead
Where Qwen3.5 4B pulls ahead
No clear win on a test scored out of 100.
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.
CodingQwen3.5 27B
- LiveCodeBench v6Qwen3.5 27B by 24.980.755.8+24.9
- SciCodeQwen3.5 27B by 23.439.516.1+23.4
- Terminal-Bench HardQwen3.5 27B by 14.432.618.2+14.4
- SWE-bench VerifiedQwen3.5 27B by 12.972.459.5+12.9
ReasoningQwen3.5 27B
- Humanity's Last ExamQwen3.5 27B by 1423.99.9+14
- GPQA DiamondQwen3.5 27B by 8.785.877.1+8.7
- CritPttie0.90tie
FactsQwen3.5 27B
- AA-Omniscience · AccuracyQwen3.5 27B by 5.620.715.1+5.6
- AA-Omniscience · Non-hallucinationQwen3.5 27B by 5.118.513.4+5.1
Images and chartsQwen3.5 27B
MathQwen3.5 27B
- HMMT Feb. 2026Qwen3.5 27B by 8.281.172.9+8.2
- AIME 2026Qwen3.5 27B by 291.789.7+2
Long documentsQwen3.5 27B
- AA-LCRQwen3.5 27B by 14.777.763+14.7
- LongBench v2Qwen3.5 27B by 10.660.650+10.6
Following instructionsQwen3.5 27B
- IFBenchQwen3.5 27B by 23.675.652+23.6
- Multi-ChallengeQwen3.5 27B by 11.860.849+11.8
Other results24 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 Agentic IndexQwen3.5 27B by 22.154.632.5+22.1
- VITA-BenchQwen3.5 27B by 19.941.922+19.9
- HallusionBenchQwen3.5 27B by 18.37051.7+18.3
- BFCLv4Qwen3.5 27B by 17.968.550.6+17.9
- RealWorldQAQwen3.5 27B by 16.683.767.1+16.6
- SimpleVQAQwen3.5 27B by 15.35640.7+15.3
- AA-OmniscienceQwen3.5 27B by 14.3-44-58.4+14.3
- ERQAQwen3.5 27B by 14.260.546.3+14.2
- HMMT 2025Qwen3.5 27B by 1389.876.8+13
- SuperGPQAQwen3.5 27B by 12.765.652.9+12.7
- Artificial Analysis Coding IndexQwen3.5 27B by 12.334.922.6+12.3
- WMT24++Qwen3.5 27B by 1177.666.6+11
- MMLU-ProXQwen3.5 27B by 10.782.271.5+10.7
- IncludeQwen3.5 27B by 10.681.671+10.6
- AA IntelligenceQwen3.5 27B by 9.822.913.1+9.8
- MMMLUQwen3.5 27B by 9.885.976.1+9.8
- MMLU-ProQwen3.5 27B by 786.179.1+7
- C-EvalQwen3.5 27B by 5.490.585.1+5.4
- IFEvalQwen3.5 27B by 5.29589.8+5.2
- DeepPlanningQwen3.5 27B by 522.617.6+5
- MMLU-ReduxQwen3.5 27B by 4.493.288.8+4.4
- Claw-Eval AvgQwen3.5 27B by 264.362.3+2
- τ²-Bench Telecom (AA run)Qwen3.5 27B by 1.893.992.1+1.8
- t2-benchtie7979.9tie
Questions people ask
Which is better, Qwen3.5 27B or Qwen3.5 4B?
Qwen3.5 27B wins all eight areas we test: coding, agents, reasoning, facts, images and charts, math, long documents and following instructions. Qwen3.5 4B wins none, but costs 93% less.
Which is better for coding?
Qwen3.5 27B. It wins 4 of the 4 coding tests both models report; Qwen3.5 4B wins none.
Which is cheaper?
Qwen3.5 27B costs $0.30 per million input tokens and $2.40 per million output tokens; Qwen3.5 4B costs $0.03 and $0.15. That makes Qwen3.5 4B about 93% 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 21 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 24 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.