Claude Opus 4.6 vs Qwen3.6 Plus
Wins 5 of 8 areas
Coding · Agents · Reasoning · Facts · Math
Wins 1 of 8 areas
Images and charts
Claude Opus 4.6 is the stronger all-rounder.Qwen3.6 Plus is cheaper and better at images and charts.
Scores updated · 60 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 software71Claude Opus 4.67 of 9 tests · 1 tie
- ReasoningHard problems that need careful thinking40Claude Opus 4.64 of 4 tests
- MathCompetition and research-level math41Claude Opus 4.64 of 5 tests
- AgentsCarrying out multi-step tasks on its own21Claude Opus 4.62 of 3 tests
- FactsGetting facts right instead of making them up21Claude Opus 4.62 of 3 tests
- Images and chartsUnderstanding pictures, charts and video02Qwen3.6 Plus2 of 2 tests
- Long documentsFinding answers in very long texts00Even0 each · 1 tie
- Following instructionsDoing exactly what it is asked11Even1 each
Long documents 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.
Qwen3.6 Plus 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 Claude Opus 4.6 pulls ahead
- Real work tasks from 44 professionsGDPVal+31.2points ahead
- Unpublished advanced math problemsFrontierMath Tiers 1-3 (v2)+27.7points ahead
- Answers hard knowledge questions correctlyAA-Omniscience · Accuracy+20.6points ahead
Where Qwen3.6 Plus pulls ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+28.2points ahead
- Follows unfamiliar, precisely checkable instructionsIFBench+22.1points ahead
- Multi-step tasks using many real software toolsMCP Atlas+11.4points ahead
Every test, side by side
All 60 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingClaude Opus 4.6
- SciCodeClaude Opus 4.6 by 11.251.940.7+11.2
- LMArena · WebDevClaude Opus 4.6 by 85 rating points15461461+85 rating
- LiveBench · Agentic CodingClaude Opus 4.6 by 7.64941.4+7.6
- SWE-bench MultilingualClaude Opus 4.6 by 477.873.8+4
- SWE-bench ProQwen3.6 Plus by 3.253.456.6+3.2
- Terminal-Bench HardClaude Opus 4.6 by 2.346.243.9+2.3
- SWE-bench VerifiedClaude Opus 4.6 by 280.878.8+2
- LiveCodeBench v6Claude Opus 4.6 by 1.788.887.1+1.7
- LiveBench · Codingtie78.278.2tie
AgentsClaude Opus 4.6
- GDPValClaude Opus 4.6 by 31.255.924.7+31.2
- MCP AtlasQwen3.6 Plus by 11.462.774.1+11.4
- OSWorld-VerifiedClaude Opus 4.6 by 10.272.762.5+10.2
ReasoningClaude Opus 4.6
- LiveBench · ReasoningClaude Opus 4.6 by 12.988.775.8+12.9
- Humanity's Last ExamClaude Opus 4.6 by 12.139.927.8+12.1
- CritPtClaude Opus 4.6 by 9.712.62.9+9.7
- GPQA DiamondClaude Opus 4.6 by 1.489.688.2+1.4
FactsClaude Opus 4.6
- AA-Omniscience · Non-hallucinationQwen3.6 Plus by 28.237.265.4+28.2
- AA-Omniscience · AccuracyClaude Opus 4.6 by 20.64726.4+20.6
- SimpleQA VerifiedClaude Opus 4.6 by 2.94744.1+2.9
Images and chartsQwen3.6 Plus
- CharXiv (RQ)Qwen3.6 Plus by 4.177.481.5+4.1
- MMMU-ProQwen3.6 Plus by 2.675.478+2.6
MathClaude Opus 4.6
- FrontierMath Tiers 1-3 (v2)Claude Opus 4.6 by 27.76638.3+27.7
- IMOAnswerBenchQwen3.6 Plus by 8.575.383.8+8.5
- HMMT Feb. 2026Claude Opus 4.6 by 8.496.287.8+8.4
- LiveBench · MathematicsClaude Opus 4.6 by 5.689.383.7+5.6
- AIME 2026Claude Opus 4.6 by 1.496.795.3+1.4
Following instructionsEven
- IFBenchQwen3.6 Plus by 22.153.175.2+22.1
- LiveBench · Instruction FollowingClaude Opus 4.6 by 563.358.3+5
Other results31 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 IndexClaude Opus 4.6 by 38.667.629+38.6
- ERQAQwen3.6 Plus by 24.940.865.7+24.9
- MATH-VisionQwen3.6 Plus by 22.565.588+22.5
- ToolathlonClaude Opus 4.6 by 1756.839.8+17
- AgentWorldBench - MCPClaude Opus 4.6 by 14.669.955.3+14.6
- frontiermath_tier_4_v1Claude Opus 4.6 by 14.622.98.3+14.6
- AA-OmniscienceClaude Opus 4.6 by 12.813.70.9+12.8
- NL2RepoClaude Opus 4.6 by 11.949.837.9+11.9
- RealWorldQAQwen3.6 Plus by 11.573.985.4+11.5
- ScreenSpot-Pro (No tools)Qwen3.6 Plus by 10.557.768.2+10.5
- AgentWorldBench - OSClaude Opus 4.6 by 9.970.260.3+9.9
- MCPMarkClaude Opus 4.6 by 8.556.748.2+8.5
- livebench_languageClaude Opus 4.6 by 8.383.375+8.3
- AgentWorldBench - SearchClaude Opus 4.6 by 7.429.321.9+7.4
- Tool-DecathlonClaude Opus 4.6 by 7.447.239.8+7.4
- AgentWorldBench - OverallClaude Opus 4.6 by 757.850.8+7
- AgentWorldBench - TerminalClaude Opus 4.6 by 6.957.550.6+6.9
- Artificial Analysis Coding IndexQwen3.6 Plus by 6.448.154.5+6.4
- τ²-Bench Telecom (AA run)Qwen3.6 Plus by 5.692.197.7+5.6
- AgentWorldBench - SWEClaude Opus 4.6 by 5.564.559.1+5.5
- LiveBenchClaude Opus 4.6 by 5.476.370.8+5.4
- AA IntelligenceClaude Opus 4.6 by 4.931.927+4.9
- AgentWorldBench - AndroidClaude Opus 4.6 by 461.757.6+4
- Terminal-Bench 2.0Claude Opus 4.6 by 3.865.461.6+3.8
- HLE (with tools)Claude Opus 4.6 by 2.753.350.6+2.7
- HMMT Nov. 2025Claude Opus 4.6 by 1.796.394.6+1.7
- τ³-BenchClaude Opus 4.6 by 1.772.470.7+1.7
- MMMLUClaude Opus 4.6 by 1.691.189.5+1.6
- AgentWorldBench - Webtie51.450.8tie
- MMLU-Protie89.188.5tie
- livebench_data_analysistie69.969.9tie
Questions people ask
Which is better, Claude Opus 4.6 or Qwen3.6 Plus?
Claude Opus 4.6 wins five of the eight areas we test: coding, agents, reasoning, facts and math. Qwen3.6 Plus wins images and charts, and costs 88% less. They are level on long documents and following instructions.
Which is better for coding?
Claude Opus 4.6. It wins 7 of the 9 coding tests both models report; Qwen3.6 Plus wins 1, and 1 is a tie.
Which is cheaper?
Claude Opus 4.6 costs $5.00 per million input tokens and $25.00 per million output tokens; Qwen3.6 Plus costs $0.50 and $3.00. That makes Qwen3.6 Plus about 88% cheaper for the same work.
How do you compare the two?
We use the 60 benchmark tests both models have published scores on. The verdict counts the 29 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 31 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.