Qwen3.5 27B vs Qwen3.6 Plus
Wins 0 of 8 areas
—
Wins 7 of 8 areas
Coding · Agents · Reasoning · Facts · Images and charts · Math · Long documents
Qwen3.6 Plus is the stronger all-rounder.Qwen3.5 27B is cheaper.
Scores updated · 61 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.
- CodingWriting and fixing software07Qwen3.6 Plus7 of 7 tests
- Images and chartsUnderstanding pictures, charts and video04Qwen3.6 Plus4 of 4 tests
- ReasoningHard problems that need careful thinking03Qwen3.6 Plus3 of 3 tests
- MathCompetition and research-level math03Qwen3.6 Plus3 of 3 tests
- FactsGetting facts right instead of making them up02Qwen3.6 Plus2 of 2 tests
- AgentsCarrying out multi-step tasks on its own12Qwen3.6 Plus2 of 3 tests
- Long documentsFinding answers in very long texts01Qwen3.6 Plus1 of 2 tests · 1 tie
- Following instructionsDoing exactly what it is asked00Even0 each · 1 tie
Following instructions 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 27B costs 23% 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 Qwen3.6 Plus pulls ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+46.9points ahead
- Hard command-line tasks in a real terminalTerminal-Bench Hard+11.3points ahead
- Harvard-MIT high-school math contest problemsHMMT Feb. 2026+6.7points ahead
Where Qwen3.5 27B pulls ahead
- Real work tasks from 44 professionsGDPVal+8.3points ahead
Every test, side by side
All 61 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingQwen3.6 Plus
- Terminal-Bench HardQwen3.6 Plus by 11.332.643.9+11.3
- LMArena · WebDevQwen3.6 Plus by 104 rating points13571461+104 rating
- LiveCodeBench v6Qwen3.6 Plus by 6.480.787.1+6.4
- SWE-bench VerifiedQwen3.6 Plus by 6.472.478.8+6.4
- SWE-bench ProQwen3.6 Plus by 5.451.256.6+5.4
- SWE-bench MultilingualQwen3.6 Plus by 4.569.373.8+4.5
- SciCodeQwen3.6 Plus by 1.239.540.7+1.2
AgentsQwen3.6 Plus
- GDPValQwen3.5 27B by 8.33324.7+8.3
- OSWorld-VerifiedQwen3.6 Plus by 6.356.262.5+6.3
- MCP AtlasQwen3.6 Plus by 5.768.474.1+5.7
ReasoningQwen3.6 Plus
- Humanity's Last ExamQwen3.6 Plus by 3.923.927.8+3.9
- GPQA DiamondQwen3.6 Plus by 2.485.888.2+2.4
- CritPtQwen3.6 Plus by 20.92.9+2
FactsQwen3.6 Plus
- AA-Omniscience · Non-hallucinationQwen3.6 Plus by 46.918.565.4+46.9
- AA-Omniscience · AccuracyQwen3.6 Plus by 5.720.726.4+5.7
Images and chartsQwen3.6 Plus
- MMMUQwen3.6 Plus by 3.782.386+3.7
- MMMU-ProQwen3.6 Plus by 37578+3
- MMStarQwen3.6 Plus by 2.38183.3+2.3
- CharXiv (RQ)Qwen3.6 Plus by 279.581.5+2
MathQwen3.6 Plus
- HMMT Feb. 2026Qwen3.6 Plus by 6.781.187.8+6.7
- IMOAnswerBenchQwen3.6 Plus by 3.979.983.8+3.9
- AIME 2026Qwen3.6 Plus by 3.691.795.3+3.6
Long documentsQwen3.6 Plus
- LongBench v2Qwen3.6 Plus by 1.460.662+1.4
- AA-LCRtie77.778.3tie
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-OmniscienceQwen3.6 Plus by 44.9-440.9+44.9
- AA Agentic IndexQwen3.5 27B by 25.654.629+25.6
- Terminal-Bench 2.0Qwen3.6 Plus by 2041.661.6+20
- Artificial Analysis Coding IndexQwen3.6 Plus by 19.634.954.5+19.6
- DeepPlanningQwen3.6 Plus by 18.922.641.5+18.9
- WideSearchQwen3.6 Plus by 13.261.174.3+13.2
- Claw Eval (pass@3)Qwen3.6 Plus by 12.546.258.7+12.5
- MCPMarkQwen3.6 Plus by 11.936.348.2+11.9
- SimpleVQAQwen3.6 Plus by 11.35667.3+11.3
- NL2RepoQwen3.6 Plus by 10.627.337.9+10.6
- Tool-DecathlonQwen3.6 Plus by 8.331.539.8+8.3
- SuperGPQAQwen3.6 Plus by 665.671.6+6
- ERQAQwen3.6 Plus by 5.260.565.7+5.2
- HMMT 2025Qwen3.6 Plus by 4.889.894.6+4.8
- HMMT Nov. 2025Qwen3.6 Plus by 4.889.894.6+4.8
- AA IntelligenceQwen3.6 Plus by 4.122.927+4.1
- τ²-Bench Telecom (AA run)Qwen3.6 Plus by 3.893.997.7+3.8
- MMMLUQwen3.6 Plus by 3.685.989.5+3.6
- IncludeQwen3.6 Plus by 3.581.685.1+3.5
- C-EvalQwen3.6 Plus by 2.890.593.3+2.8
- MMLU-ProXQwen3.6 Plus by 2.582.284.7+2.5
- CC-OCRQwen3.6 Plus by 2.48183.4+2.4
- MMLU-ProQwen3.6 Plus by 2.486.188.5+2.4
- VITA-BenchQwen3.6 Plus by 2.441.944.3+2.4
- OmniDocBench 1.5Qwen3.6 Plus by 2.388.991.2+2.3
- τ³-BenchQwen3.6 Plus by 2.368.470.7+2.3
- ScreenSpot-Pro (No tools)Qwen3.5 27B by 2.170.368.2+2.1
- MATH-VisionQwen3.6 Plus by 28688+2
- MMlongBench DocQwen3.6 Plus by 1.860.262+1.8
- RealWorldQAQwen3.6 Plus by 1.783.785.4+1.7
- VideoMMMUQwen3.6 Plus by 1.782.384+1.7
- AI2DQwen3.6 Plus by 1.592.994.4+1.5
- MMLU-ReduxQwen3.6 Plus by 1.393.294.5+1.3
- MLVUtie85.986.7tie
- IFEvaltie9594.3tie
- DynaMathtie87.788tie
Questions people ask
Which is better, Qwen3.5 27B or Qwen3.6 Plus?
Qwen3.6 Plus wins seven of the eight areas we test: coding, agents, reasoning, facts, images and charts, math and long documents. Qwen3.5 27B wins none, but costs 23% less. They are level on following instructions.
Which is better for coding?
Qwen3.6 Plus. It wins 7 of the 7 coding tests both models report; Qwen3.5 27B wins none.
Which is cheaper?
Qwen3.5 27B costs $0.30 per million input tokens and $2.40 per million output tokens; Qwen3.6 Plus costs $0.50 and $3.00. That makes Qwen3.5 27B about 23% cheaper for the same work.
How do you compare the two?
We use the 61 benchmark tests both models have published scores on. The verdict counts the 25 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.