Claude Sonnet 4.5 vs Qwen3.5 35B A3B
Wins 4 of 8 areas
Coding · Agents · Facts · Long documents
Wins 4 of 8 areas
Reasoning · Images and charts · Math · Following instructions
The two are evenly matched.Claude Sonnet 4.5 is better at coding and agents; Qwen3.5 35B A3B at images and charts and reasoning.
Scores updated · 52 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 video04Qwen3.5 35B A3B4 of 4 tests
- ReasoningHard problems that need careful thinking02Qwen3.5 35B A3B2 of 3 tests · 1 tie
- Following instructionsDoing exactly what it is asked02Qwen3.5 35B A3B2 of 2 tests
- MathCompetition and research-level math01Qwen3.5 35B A3B1 of 1 test
- CodingWriting and fixing software62Claude Sonnet 4.56 of 8 tests
- AgentsCarrying out multi-step tasks on its own32Claude Sonnet 4.53 of 5 tests
- FactsGetting facts right instead of making them up21Claude Sonnet 4.52 of 3 tests
- Long documentsFinding answers in very long texts10Claude Sonnet 4.51 of 2 tests · 1 tie
Math 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 35B A3B costs 88% less for the same work.
The biggest differences
The three tests each model wins by the widest margin. Scores are out of 100.
Where Qwen3.5 35B A3B pulls ahead
- Finds hard-to-locate facts by browsing the webBrowseComp+36.9points ahead
- Multi-step tasks using many real software toolsMCP Atlas+18.6points ahead
- Follows unfamiliar, precisely checkable instructionsIFBench+15.2points ahead
Where Claude Sonnet 4.5 pulls ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+36.3points ahead
- Customer-service tasks in a simulated bankτ-Bench V3 · Banking+19.6points ahead
- Command-line tasks in a real terminalTerminal-Bench 2.1+15points ahead
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.
CodingClaude Sonnet 4.5
- Terminal-Bench 2.1Claude Sonnet 4.5 by 1555.840.8+15
- LMArena · WebDevClaude Sonnet 4.5 by 143 rating points13931250+143 rating
- LiveCodeBench v6Qwen3.5 35B A3B by 10.66474.6+10.6
- Terminal-Bench HardClaude Sonnet 4.5 by 9.135.626.5+9.1
- SciCodeClaude Sonnet 4.5 by 845.737.7+8
- SWE-bench VerifiedClaude Sonnet 4.5 by 877.269.2+8
- SWE-bench MultilingualClaude Sonnet 4.5 by 6.76760.3+6.7
- SWE-bench ProQwen3.5 35B A3B by 143.644.6+1
AgentsClaude Sonnet 4.5
- BrowseCompQwen3.5 35B A3B by 36.924.161+36.9
- τ-Bench V3 · BankingClaude Sonnet 4.5 by 19.624.54.9+19.6
- MCP AtlasQwen3.5 35B A3B by 18.643.862.4+18.6
- GDPValClaude Sonnet 4.5 by 14.920.55.6+14.9
- OSWorld-VerifiedClaude Sonnet 4.5 by 6.961.454.5+6.9
ReasoningQwen3.5 35B A3B
- Humanity's Last ExamQwen3.5 35B A3B by 3.217.821+3.2
- GPQA DiamondQwen3.5 35B A3B by 1.183.484.5+1.1
- CritPttie1.10.9tie
FactsClaude Sonnet 4.5
- AA-Omniscience · Non-hallucinationClaude Sonnet 4.5 by 36.350.914.6+36.3
- AA-Omniscience · AccuracyClaude Sonnet 4.5 by 12.832.920.1+12.8
- Vectara HHEM hallucination ratelower is betterQwen3.5 35B A3B by 1.51210.5+1.5
Images and chartsQwen3.5 35B A3B
- CharXiv (RQ)Qwen3.5 35B A3B by 10.367.277.5+10.3
- MathVistaQwen3.5 35B A3B by 6.479.886.2+6.4
- MMMU-ProQwen3.5 35B A3B by 468.772.7+4
- MMMUQwen3.5 35B A3B by 1.879.681.4+1.8
Long documentsClaude Sonnet 4.5
- LongBench v2Claude Sonnet 4.5 by 2.861.859+2.8
- AA-LCRtie72.372tie
Following instructionsQwen3.5 35B A3B
- IFBenchQwen3.5 35B A3B by 15.257.372.5+15.2
- Multi-ChallengeQwen3.5 35B A3B by 4.755.360+4.7
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-OmniscienceClaude Sonnet 4.5 by 48-0.1-48.1+48
- BrowseComp-ZHQwen3.5 35B A3B by 27.142.469.5+27.1
- HMMT Feb. 2025Qwen3.5 35B A3B by 21.567.589+21.5
- Artificial Analysis Coding IndexClaude Sonnet 4.5 by 15.152.137+15.1
- HMMT 2025Qwen3.5 35B A3B by 14.674.689.2+14.6
- RealWorldQAQwen3.5 35B A3B by 13.870.384.1+13.8
- CC-OCRQwen3.5 35B A3B by 12.668.180.7+12.6
- τ²-Bench Telecom (AA run)Qwen3.5 35B A3B by 11.178.189.2+11.1
- Terminal-Bench 2.0Claude Sonnet 4.5 by 9.55040.5+9.5
- HallusionBenchQwen3.5 35B A3B by 859.967.9+8
- HMMT Nov. 2025Qwen3.5 35B A3B by 7.581.789.2+7.5
- AA Agentic IndexClaude Sonnet 4.5 by 5.717.511.8+5.7
- AI2DQwen3.5 35B A3B by 5.68792.6+5.6
- vectara_answer_rateQwen3.5 35B A3B by 4.295.699.8+4.2
- MMMLUClaude Sonnet 4.5 by 3.989.185.2+3.9
- OmniDocBench 1.5Qwen3.5 35B A3B by 3.585.889.3+3.5
- vectara_avg_summary_lengthClaude Sonnet 4.5 by 32.9 rating points127.894.9+32.9 rating
- MMBench EN-DEV-v1.1Qwen3.5 35B A3B by 3.288.391.5+3.2
- MMLU-ReduxClaude Sonnet 4.5 by 2.395.693.3+2.3
- MMLU-ProClaude Sonnet 4.5 by 2.287.585.3+2.2
- vectara_factual_consistencyQwen3.5 35B A3B by 1.58889.5+1.5
- AA IntelligenceClaude Sonnet 4.5 by 1.420.719.3+1.4
- GPQA (unspecified)tie83.484.2tie
- SimpleVQAtie57.658.3tie
Questions people ask
Which is better, Claude Sonnet 4.5 or Qwen3.5 35B A3B?
Claude Sonnet 4.5 and Qwen3.5 35B A3B each win four of the eight areas we test. Claude Sonnet 4.5 wins coding, agents, facts and long documents; Qwen3.5 35B A3B wins reasoning, images and charts, math and following instructions. Qwen3.5 35B A3B costs 88% less.
Which is better for coding?
Claude Sonnet 4.5. It wins 6 of the 8 coding tests both models report; Qwen3.5 35B A3B wins 2.
Which is cheaper?
Claude Sonnet 4.5 costs $3.00 per million input tokens and $15.00 per million output tokens; Qwen3.5 35B A3B costs $0.25 and $2.00. That makes Qwen3.5 35B A3B about 88% cheaper for the same work.
How do you compare the two?
We use the 52 benchmark tests both models have published scores on. The verdict counts the 28 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.