Claude Opus 4.6 vs Claude Sonnet 4.6
Wins 5 of 8 areas
Coding · Agents · Reasoning · Images and charts · Math
Wins 2 of 8 areas
Long documents · Following instructions
Claude Opus 4.6 is the stronger all-rounder.Claude Sonnet 4.6 is cheaper and better at long documents.
Scores updated · 66 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.
- ReasoningHard problems that need careful thinking50Claude Opus 4.65 of 5 tests
- AgentsCarrying out multi-step tasks on its own40Claude Opus 4.64 of 5 tests · 1 tie
- Images and chartsUnderstanding pictures, charts and video30Claude Opus 4.63 of 3 tests
- MathCompetition and research-level math20Claude Opus 4.62 of 2 tests
- CodingWriting and fixing software43Claude Opus 4.64 of 7 tests
- Long documentsFinding answers in very long texts01Claude Sonnet 4.61 of 2 tests · 1 tie
- Following instructionsDoing exactly what it is asked01Claude Sonnet 4.61 of 2 tests · 1 tie
- FactsGetting facts right instead of making them up22Even2 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.
Claude Sonnet 4.6 costs 40% 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+19.2points ahead
- Short factual questions, answered correctlySimpleQA Verified+14.2points ahead
- Proof problems from the US Math OlympiadUSAMO 2026+11.2points ahead
Where Claude Sonnet 4.6 pulls ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+14.4points ahead
- Hard command-line tasks in a real terminalTerminal-Bench Hard+6.8points ahead
- Long, multi-file coding tasks in real codebasesSWE-bench Pro+4.7points ahead
Every test, side by side
All 66 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingClaude Opus 4.6
- Terminal-Bench HardClaude Sonnet 4.6 by 6.846.253+6.8
- LiveBench · Agentic CodingClaude Opus 4.6 by 6.44942.6+6.4
- SWE-bench ProClaude Sonnet 4.6 by 4.753.458.1+4.7
- LMArena · WebDevClaude Opus 4.6 by 24 rating points15461522+24 rating
- SciCodeClaude Opus 4.6 by 1.851.950.1+1.8
- SWE-bench VerifiedClaude Opus 4.6 by 1.280.879.6+1.2
- LiveBench · CodingClaude Sonnet 4.6 by 1.178.279.3+1.1
AgentsClaude Opus 4.6
- GDPValClaude Opus 4.6 by 19.255.936.7+19.2
- BrowseCompClaude Opus 4.6 by 9.38474.7+9.3
- AA ApexAgentsClaude Opus 4.6 by 53328+5
- MCP AtlasClaude Opus 4.6 by 1.462.761.3+1.4
- OSWorld-Verifiedtie72.772.5tie
ReasoningClaude Opus 4.6
- ARC-AGI-2Claude Opus 4.6 by 10.568.858.3+10.5
- CritPtClaude Opus 4.6 by 9.512.63.1+9.5
- Humanity's Last ExamClaude Opus 4.6 by 6.339.933.6+6.3
- LiveBench · ReasoningClaude Opus 4.6 by 3.988.784.8+3.9
- GPQA DiamondClaude Opus 4.6 by 2.189.687.5+2.1
FactsEven
- AA-Omniscience · Non-hallucinationClaude Sonnet 4.6 by 14.437.251.6+14.4
- SimpleQA VerifiedClaude Opus 4.6 by 14.24732.8+14.2
- AA-Omniscience · AccuracyClaude Opus 4.6 by 6.14740.9+6.1
- Vectara HHEM hallucination ratelower is betterClaude Sonnet 4.6 by 1.612.210.6+1.6
Images and chartsClaude Opus 4.6
- CharXiv (RQ)Claude Opus 4.6 by 5.877.471.6+5.8
- LMArena · VisionClaude Opus 4.6 by 33 rating points13161283+33 rating
- MMMU-ProClaude Opus 4.6 by 2.175.473.3+2.1
MathClaude Opus 4.6
- USAMO 2026Claude Opus 4.6 by 11.266.255+11.2
- LiveBench · MathematicsClaude Opus 4.6 by 2.389.387+2.3
Long documentsClaude Sonnet 4.6
- AA-LCRClaude Sonnet 4.6 by 27880+2
- MRCR v2 8 needle 128k (average)tie8484.9tie
Following instructionsClaude Sonnet 4.6
- IFBenchClaude Sonnet 4.6 by 3.553.156.6+3.5
- LiveBench · Instruction Followingtie63.363.2tie
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 Agentic IndexClaude Opus 4.6 by 34.567.633.1+34.5
- τ²-Bench Telecom (AA run)Claude Opus 4.6 by 16.492.175.7+16.4
- FLTEvalClaude Opus 4.6 by 15.939.623.7+15.9
- ToolathlonClaude Opus 4.6 by 15.856.841+15.8
- Artificial Analysis Coding IndexClaude Sonnet 4.6 by 14.948.163+14.9
- frontiermath_tier_4_v1Claude Opus 4.6 by 14.622.98.3+14.6
- Toolathlon Pass@∞Claude Opus 4.6 by 9.247.238+9.2
- livebench_data_analysisClaude Sonnet 4.6 by 8.169.978+8.1
- BrowseComp (single-agent)Claude Opus 4.6 by 7.583.776.2+7.5
- livebench_languageClaude Opus 4.6 by 7.283.376.1+7.2
- AgentWorldBench - OSClaude Opus 4.6 by 770.263.2+7
- ARC-AGI-1Claude Opus 4.6 by 79386+7
- HLE (with tools)Claude Opus 4.6 by 6.553.346.8+6.5
- Terminal-Bench 2.0Claude Opus 4.6 by 6.365.459.1+6.3
- GDPval-AA (Elo)Claude Sonnet 4.6 by 57 rating points16191676+57 rating
- OfficeQAClaude Opus 4.6 by 4.873.568.7+4.8
- AgentWorldBench - AndroidClaude Opus 4.6 by 3.761.758+3.7
- OfficeQA ProClaude Opus 4.6 by 3.757.153.4+3.7
- Finance AgentClaude Sonnet 4.6 by 2.660.763.3+2.6
- vectara_avg_summary_lengthClaude Opus 4.6 by 22.9 rating points137.6114.7+22.9 rating
- AA IntelligenceClaude Opus 4.6 by 1.831.930.1+1.8
- AgentWorldBench - OverallClaude Opus 4.6 by 1.857.856+1.8
- MMMLUClaude Opus 4.6 by 1.891.189.3+1.8
- vectara_factual_consistencyClaude Sonnet 4.6 by 1.687.889.4+1.6
- AA-OmniscienceClaude Opus 4.6 by 1.513.712.2+1.5
- Legal Agent BenchmarkClaude Sonnet 4.6 by 1.24.25.4+1.2
- LiveBenchtie76.375.5tie
- AgentWorldBench - Webtie51.450.8tie
- CharXiv Reasoning (With tools)tie84.785.3tie
- AgentWorldBench - Searchtie29.328.8tie
- AgentWorldBench - Terminaltie57.557tie
- Toolathlon Avg turnstie16.916.5tie
- τ²-Bench (Retail)tie91.991.7tie
- AgentWorldBench - MCPtie69.970tie
- AgentWorldBench - SWEtie64.564.5tie
- vectara_answer_ratetie99.899.9tie
Questions people ask
Which is better, Claude Opus 4.6 or Claude Sonnet 4.6?
Claude Opus 4.6 wins five of the eight areas we test: coding, agents, reasoning, images and charts and math. Claude Sonnet 4.6 wins long documents and following instructions, and costs 40% less. They are level on facts.
Which is better for coding?
Claude Opus 4.6. It wins 4 of the 7 coding tests both models report; Claude Sonnet 4.6 wins 3.
Which is cheaper?
Claude Opus 4.6 costs $5.00 per million input tokens and $25.00 per million output tokens; Claude Sonnet 4.6 costs $3.00 and $15.00. That makes Claude Sonnet 4.6 about 40% cheaper for the same work.
How do you compare the two?
We use the 66 benchmark tests both models have published scores on. The verdict counts the 30 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.