Claude Opus 4.5 vs Claude Sonnet 4.5
Wins 8 of 8 areas
Coding · Agents · Reasoning · Facts · Images and charts · Math · Long documents · Following instructions
Wins 0 of 8 areas
—
Claude Opus 4.5 is the stronger all-rounder.Claude Sonnet 4.5 is cheaper.
Scores updated · 73 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 software61Claude Opus 4.56 of 7 tests
- AgentsCarrying out multi-step tasks on its own40Claude Opus 4.54 of 4 tests
- ReasoningHard problems that need careful thinking40Claude Opus 4.54 of 4 tests
- MathCompetition and research-level math30Claude Opus 4.53 of 3 tests
- FactsGetting facts right instead of making them up31Claude Opus 4.53 of 4 tests
- Images and chartsUnderstanding pictures, charts and video20Claude Opus 4.52 of 4 tests · 2 ties
- Long documentsFinding answers in very long texts20Claude Opus 4.52 of 2 tests
- Following instructionsDoing exactly what it is asked10Claude Opus 4.51 of 2 tests · 1 tie
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 4.5 costs 40% 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 Claude Opus 4.5 pulls ahead
- Real work tasks from 44 professionsGDPVal+26.8points ahead
- Abstract visual puzzles that people can solveARC-AGI-2+24points ahead
- Recent programming contest problemsLiveCodeBench v6+20.8points ahead
Where Claude Sonnet 4.5 pulls ahead
- Fixes real GitHub issues in many programming languagesSWE-bench Multilingual+60points ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+11.9points ahead
Every test, side by side
All 73 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingClaude Opus 4.5
- SWE-bench MultilingualClaude Sonnet 4.5 by 60767+60
- LiveCodeBench v6Claude Opus 4.5 by 20.884.864+20.8
- Terminal-Bench HardClaude Opus 4.5 by 11.44735.6+11.4
- SWE-bench ProClaude Opus 4.5 by 8.45243.6+8.4
- LMArena · WebDevClaude Opus 4.5 by 75 rating points14681393+75 rating
- SciCodeClaude Opus 4.5 by 3.849.545.7+3.8
- SWE-bench VerifiedClaude Opus 4.5 by 3.780.977.2+3.7
AgentsClaude Opus 4.5
- GDPValClaude Opus 4.5 by 26.847.320.5+26.8
- MCP AtlasClaude Opus 4.5 by 18.562.343.8+18.5
- BrowseCompClaude Opus 4.5 by 12.93724.1+12.9
- OSWorld-VerifiedClaude Opus 4.5 by 4.966.361.4+4.9
ReasoningClaude Opus 4.5
- ARC-AGI-2Claude Opus 4.5 by 2437.613.6+24
- Humanity's Last ExamClaude Opus 4.5 by 12.330.117.8+12.3
- CritPtClaude Opus 4.5 by 3.54.61.1+3.5
- GPQA DiamondClaude Opus 4.5 by 3.286.683.4+3.2
FactsClaude Opus 4.5
- SimpleQA VerifiedClaude Opus 4.5 by 1545.730.7+15
- AA-Omniscience · AccuracyClaude Opus 4.5 by 13.746.632.9+13.7
- AA-Omniscience · Non-hallucinationClaude Sonnet 4.5 by 11.93950.9+11.9
- Vectara HHEM hallucination ratelower is betterClaude Opus 4.5 by 1.110.912+1.1
Images and chartsClaude Opus 4.5
- MMMU-ProClaude Opus 4.5 by 5.37468.7+5.3
- MMMUClaude Opus 4.5 by 1.180.779.6+1.1
- MathVistatie80.279.8tie
- CharXiv (RQ)tie67.267.2tie
MathClaude Opus 4.5
- IMOAnswerBenchClaude Opus 4.5 by 18.18465.9+18.1
- FrontierMath Tiers 1-3 (v2)Claude Opus 4.5 by 10.534.423.9+10.5
- FrontierMath Tier 4Claude Opus 4.5 by 2.54.92.4+2.5
Long documentsClaude Opus 4.5
- AA-LCRClaude Opus 4.5 by 577.372.3+5
- LongBench v2Claude Opus 4.5 by 2.664.461.8+2.6
Following instructionsClaude Opus 4.5
- Multi-ChallengeClaude Opus 4.5 by 3.75955.3+3.7
- IFBenchtie5857.3tie
Other results43 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 4.5 by 42.159.617.5+42.1
- BrowseComp (context management)Claude Opus 4.5 by 31.757.826.1+31.7
- HMMT Feb. 2025Claude Opus 4.5 by 25.492.967.5+25.4
- OJ-Bench (cpp)Claude Opus 4.5 by 24.254.630.4+24.2
- ARC-AGI-2 (Verified)Claude Opus 4.5 by 2437.613.6+24
- ARC-AGI-1Claude Opus 4.5 by 18.38263.7+18.3
- LiveCodeBenchClaude Opus 4.5 by 168771+16
- AA-OmniscienceClaude Opus 4.5 by 14.114-0.1+14.1
- OctoCodingbenchClaude Opus 4.5 by 13.436.222.8+13.4
- SimpleVQAClaude Opus 4.5 by 12.169.757.6+12.1
- HMMT Nov. 2025Claude Opus 4.5 by 11.693.381.7+11.6
- τ²-Bench Telecom (AA run)Claude Opus 4.5 by 11.489.578.1+11.4
- HLE (with tools)Claude Opus 4.5 by 11.243.232+11.2
- SWT-benchClaude Opus 4.5 by 10.780.269.5+10.7
- Terminal-Bench 2.0Claude Opus 4.5 by 9.359.350+9.3
- VIBE-iOSClaude Opus 4.5 by 8.89081.2+8.8
- AA IntelligenceClaude Opus 4.5 by 8.429.120.7+8.4
- BrowseComp-Plus (with tool result clearing)Claude Opus 4.5 by 7.267.660.4+7.2
- VIBE-BackendClaude Opus 4.5 by 7.29890.8+7.2
- AIME 2025Claude Opus 4.5 by 5.892.887+5.8
- BrowseComp-Plus (with tool result clearing + memory)Claude Opus 4.5 by 5.772.967.2+5.7
- Multi-SWE-BenchClaude Opus 4.5 by 5.75044.3+5.7
- LAB-Bench FigQA (with tools)Claude Opus 4.5 by 5.569.263.7+5.5
- VIBE (Average)Claude Opus 4.5 by 5.590.785.2+5.5
- VIBE-SimulationClaude Opus 4.5 by 4.98479.1+4.9
- VIBE-AndroidClaude Opus 4.5 by 4.792.287.5+4.7
- Artificial Analysis Coding IndexClaude Sonnet 4.5 by 4.347.852.1+4.3
- vectara_answer_rateClaude Opus 4.5 by 3.198.795.6+3.1
- AIME25 no toolsClaude Opus 4.5 by 39188+3
- swe_bench_bashClaude Opus 4.5 by 374.471.4+3
- τ²-Bench (Retail)Claude Opus 4.5 by 2.788.986.2+2.7
- LAB-Bench FigQA (no tools)Claude Opus 4.5 by 2.654.952.3+2.6
- ToolathlonClaude Opus 4.5 by 2.543.541+2.5
- MMLU-ProClaude Opus 4.5 by 289.587.5+2
- OmniDocBench 1.5Claude Opus 4.5 by 1.987.785.8+1.9
- VIBE-WebClaude Opus 4.5 by 1.889.187.3+1.8
- MMMLUClaude Opus 4.5 by 1.790.889.1+1.7
- SWE-PerfClaude Opus 4.5 by 1.74.73+1.7
- vectara_avg_summary_lengthClaude Sonnet 4.5 by 13.3 rating points114.5127.8+13.3 rating
- vectara_factual_consistencyClaude Opus 4.5 by 1.189.188+1.1
- τ²-Benchtie98.298tie
- frontiermath_tier_4_v1tie4.24.2tie
- MMLU-Reduxtie95.695.6tie
Questions people ask
Which is better, Claude Opus 4.5 or Claude Sonnet 4.5?
Claude Opus 4.5 wins all eight areas we test: coding, agents, reasoning, facts, images and charts, math, long documents and following instructions. Claude Sonnet 4.5 wins none, but costs 40% less.
Which is better for coding?
Claude Opus 4.5. It wins 6 of the 7 coding tests both models report; Claude Sonnet 4.5 wins 1.
Which is cheaper?
Claude Opus 4.5 costs $5.00 per million input tokens and $25.00 per million output tokens; Claude Sonnet 4.5 costs $3.00 and $15.00. That makes Claude Sonnet 4.5 about 40% cheaper for the same work.
How do you compare the two?
We use the 73 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 43 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.