Claude Sonnet 4.5 vs Qwen3.6 35B A3B
Wins 2 of 8 areas
Coding · Facts
Wins 4 of 8 areas
Agents · Reasoning · Images and charts · Following instructions
Qwen3.6 35B A3B wins more areas, narrowly.Claude Sonnet 4.5 is better at facts.
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 video04Qwen3.6 35B A3B4 of 4 tests
- AgentsCarrying out multi-step tasks on its own12Qwen3.6 35B A3B2 of 5 tests · 2 ties
- ReasoningHard problems that need careful thinking01Qwen3.6 35B A3B1 of 3 tests · 2 ties
- Following instructionsDoing exactly what it is asked01Qwen3.6 35B A3B1 of 1 test
- FactsGetting facts right instead of making them up20Claude Sonnet 4.52 of 2 tests
- CodingWriting and fixing software32Claude Sonnet 4.53 of 7 tests · 2 ties
- MathCompetition and research-level math11Even1 each
- Long documentsFinding answers in very long texts00Even0 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.
Qwen3.6 35B A3B costs 85% less for the same work.
The biggest differences
The three tests each model wins by the widest margin. Scores are out of 100.
Where Qwen3.6 35B A3B pulls ahead
- Finds hard-to-locate facts by browsing the webBrowseComp+24.6points ahead
- Multi-step tasks using many real software toolsMCP Atlas+19points ahead
- Recent programming contest problemsLiveCodeBench v6+16.4points ahead
Where Claude Sonnet 4.5 pulls ahead
- Customer-service tasks in a simulated bankτ-Bench V3 · Banking+15.2points ahead
- Answers hard knowledge questions correctlyAA-Omniscience · Accuracy+14.1points ahead
- Command-line tasks in a real terminalTerminal-Bench 2.1+10.9points ahead
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.
CodingClaude Sonnet 4.5
- LiveCodeBench v6Qwen3.6 35B A3B by 16.46480.4+16.4
- Terminal-Bench 2.1Claude Sonnet 4.5 by 10.955.844.9+10.9
- SciCodeClaude Sonnet 4.5 by 9.145.736.6+9.1
- SWE-bench ProQwen3.6 35B A3B by 5.943.649.5+5.9
- SWE-bench VerifiedClaude Sonnet 4.5 by 3.877.273.4+3.8
- Terminal-Bench Hardtie35.634.8tie
- SWE-bench Multilingualtie6767.2tie
AgentsQwen3.6 35B A3B
- BrowseCompQwen3.6 35B A3B by 24.624.148.7+24.6
- MCP AtlasQwen3.6 35B A3B by 1943.862.8+19
- τ-Bench V3 · BankingClaude Sonnet 4.5 by 15.224.59.3+15.2
- GDPValtie20.519.8tie
- Terminal-Bench 4.0tie00tie
ReasoningQwen3.6 35B A3B
- Humanity's Last ExamQwen3.6 35B A3B by 4.417.822.2+4.4
- CritPttie1.10.3tie
- GPQA Diamondtie83.484.1tie
FactsClaude Sonnet 4.5
- AA-Omniscience · AccuracyClaude Sonnet 4.5 by 14.132.918.8+14.1
- AA-Omniscience · Non-hallucinationClaude Sonnet 4.5 by 1.450.949.5+1.4
Images and chartsQwen3.6 35B A3B
- CharXiv (RQ)Qwen3.6 35B A3B by 10.867.278+10.8
- MathVistaQwen3.6 35B A3B by 6.679.886.4+6.6
- MMMU-ProQwen3.6 35B A3B by 6.368.775+6.3
- MMMUQwen3.6 35B A3B by 2.179.681.7+2.1
MathEven
- IMOAnswerBenchQwen3.6 35B A3B by 1365.978.9+13
- FrontierMath Tiers 1-3 (v2)Claude Sonnet 4.5 by 3.523.920.4+3.5
Following instructionsQwen3.6 35B A3B
- IFBenchQwen3.6 35B A3B by 7.157.364.4+7.1
Other results20 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 Feb. 2025Qwen3.6 35B A3B by 23.267.590.7+23.2
- AA-OmniscienceClaude Sonnet 4.5 by 22.1-0.1-22.2+22.1
- τ²-Bench Telecom (AA run)Qwen3.6 35B A3B by 17.278.195.3+17.2
- RealWorldQAQwen3.6 35B A3B by 1570.385.3+15
- HMMT 2025Qwen3.6 35B A3B by 14.574.689.1+14.5
- ToolathlonClaude Sonnet 4.5 by 14.14126.9+14.1
- CC-OCRQwen3.6 35B A3B by 13.868.181.9+13.8
- Artificial Analysis Coding IndexClaude Sonnet 4.5 by 10.252.141.9+10.2
- HallusionBenchQwen3.6 35B A3B by 9.959.969.8+9.9
- HMMT Nov. 2025Qwen3.6 35B A3B by 7.481.789.1+7.4
- AI2DQwen3.6 35B A3B by 5.78792.7+5.7
- MMBench EN-DEV-v1.1Qwen3.6 35B A3B by 4.588.392.8+4.5
- OmniDocBench 1.5Qwen3.6 35B A3B by 4.185.889.9+4.1
- GPQA (unspecified)Qwen3.6 35B A3B by 2.683.486+2.6
- AA Agentic IndexClaude Sonnet 4.5 by 2.517.515+2.5
- AA IntelligenceClaude Sonnet 4.5 by 2.520.718.2+2.5
- MMLU-ProClaude Sonnet 4.5 by 2.387.585.2+2.3
- MMLU-ReduxClaude Sonnet 4.5 by 2.395.693.3+2.3
- Terminal-Bench 2.0Qwen3.6 35B A3B by 1.55051.5+1.5
- SimpleVQAQwen3.6 35B A3B by 1.357.658.9+1.3
Questions people ask
Which is better, Claude Sonnet 4.5 or Qwen3.6 35B A3B?
Qwen3.6 35B A3B wins four of the eight areas we test: agents, reasoning, images and charts and following instructions. Claude Sonnet 4.5 wins coding and facts. They are level on math and long documents.
Which is better for coding?
Claude Sonnet 4.5. It wins 3 of the 7 coding tests both models report; Qwen3.6 35B A3B wins 2, and 2 are ties.
Which is cheaper?
Claude Sonnet 4.5 costs $3.00 per million input tokens and $15.00 per million output tokens; Qwen3.6 35B A3B costs $0.38 and $2.25. That makes Qwen3.6 35B A3B about 85% 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 25 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 20 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.