Claude Opus 4.5 vs Claude Opus 4.7
Wins 0 of 8 areas
—
Wins 8 of 8 areas
Coding · Agents · Reasoning · Facts · Images and charts · Math · Long documents · Following instructions
Claude Opus 4.7 is the stronger all-rounder.
Scores updated · 53 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 software08Claude Opus 4.78 of 8 tests
- ReasoningHard problems that need careful thinking05Claude Opus 4.75 of 5 tests
- MathCompetition and research-level math04Claude Opus 4.74 of 5 tests · 1 tie
- Images and chartsUnderstanding pictures, charts and video03Claude Opus 4.73 of 3 tests
- AgentsCarrying out multi-step tasks on its own13Claude Opus 4.73 of 4 tests
- FactsGetting facts right instead of making them up13Claude Opus 4.73 of 4 tests
- Following instructionsDoing exactly what it is asked01Claude Opus 4.71 of 2 tests · 1 tie
- Long documentsFinding answers in very long texts01Claude Opus 4.71 of 1 test
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.
The biggest differences
The tests each model wins by the widest margin, up to three each. Scores are out of 100.
Where Claude Opus 4.7 pulls ahead
- Fixes real GitHub issues in many programming languagesSWE-bench Multilingual+73.5points ahead
- Finds hard-to-locate facts by browsing the webBrowseComp+42.3points ahead
- Abstract visual puzzles that people can solveARC-AGI-2+38.2points ahead
Where Claude Opus 4.5 pulls ahead
- Real work tasks from 44 professionsGDPVal+4.5points ahead
Every test, side by side
All 53 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingClaude Opus 4.7
- SWE-bench MultilingualClaude Opus 4.7 by 73.5780.5+73.5
- SWE-bench ProClaude Opus 4.7 by 12.35264.3+12.3
- LiveBench · Agentic CodingClaude Opus 4.7 by 1139.750.7+11
- LMArena · WebDevClaude Opus 4.7 by 89 rating points14681557+89 rating
- SWE-bench VerifiedClaude Opus 4.7 by 6.780.987.6+6.7
- SciCodeClaude Opus 4.7 by 549.554.5+5
- Terminal-Bench HardClaude Opus 4.7 by 4.54751.5+4.5
- LiveBench · CodingClaude Opus 4.7 by 2.479.782.1+2.4
AgentsClaude Opus 4.7
- BrowseCompClaude Opus 4.7 by 42.33779.3+42.3
- MCP AtlasClaude Opus 4.7 by 1562.377.3+15
- OSWorld-VerifiedClaude Opus 4.7 by 11.766.378+11.7
- GDPValClaude Opus 4.5 by 4.547.342.8+4.5
ReasoningClaude Opus 4.7
- ARC-AGI-2Claude Opus 4.7 by 38.237.675.8+38.2
- Humanity's Last ExamClaude Opus 4.7 by 12.230.142.3+12.2
- CritPtClaude Opus 4.7 by 7.44.612+7.4
- LiveBench · ReasoningClaude Opus 4.7 by 7.180.187.2+7.1
- GPQA DiamondClaude Opus 4.7 by 4.886.691.4+4.8
FactsClaude Opus 4.7
- AA-Omniscience · Non-hallucinationClaude Opus 4.7 by 18.73957.7+18.7
- SimpleQA VerifiedClaude Opus 4.7 by 645.751.7+6
- AA-Omniscience · AccuracyClaude Opus 4.7 by 2.346.648.9+2.3
- Vectara HHEM hallucination ratelower is betterClaude Opus 4.5 by 1.110.912+1.1
Images and chartsClaude Opus 4.7
- CharXiv (RQ)Claude Opus 4.7 by 23.867.291+23.8
- MMMU-ProClaude Opus 4.7 by 4.87478.8+4.8
- MathVistaClaude Opus 4.7 by 4.280.284.4+4.2
MathClaude Opus 4.7
- FrontierMath Tiers 1-3 (v2)Claude Opus 4.7 by 35.834.470.2+35.8
- FrontierMath Tier 4Claude Opus 4.7 by 26.84.931.7+26.8
- HMMT Feb. 2026Claude Opus 4.7 by 8.685.393.9+8.6
- LiveBench · MathematicsClaude Opus 4.7 by 2.590.492.8+2.5
- AIME 2026tie95.195.8tie
Following instructionsClaude Opus 4.7
- LiveBench · Instruction FollowingClaude Opus 4.7 by 4.162.566.7+4.1
- IFBenchtie5858.6tie
Other results21 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.
- Artificial Analysis Coding IndexClaude Opus 4.7 by 25.847.873.6+25.8
- CyberGymClaude Opus 4.7 by 22.550.673.1+22.5
- AA Agentic IndexClaude Opus 4.5 by 20.159.639.5+20.1
- frontiermath_tier_4_v1Claude Opus 4.7 by 18.74.222.9+18.7
- ToolathlonClaude Opus 4.7 by 15.843.559.3+15.8
- AA-OmniscienceClaude Opus 4.7 by 13.31427.3+13.3
- SimpleVQAClaude Opus 4.5 by 13.269.756.5+13.2
- AA IntelligenceClaude Opus 4.7 by 11.629.140.7+11.6
- HLE (with tools)Claude Opus 4.7 by 11.543.254.7+11.5
- Terminal-Bench 2.0Claude Opus 4.7 by 10.159.369.4+10.1
- ARC-AGI-1Claude Opus 4.7 by 108292+10
- Finance AgentClaude Opus 4.7 by 9.255.264.4+9.2
- livebench_data_analysisClaude Opus 4.7 by 3.974.478.3+3.9
- vectara_avg_summary_lengthClaude Opus 4.7 by 34.6 rating points114.5149.1+34.6 rating
- livebench_languageClaude Opus 4.5 by 3.481.377.9+3.4
- vectara_factual_consistencyClaude Opus 4.5 by 1.189.188+1.1
- LiveBenchtie7676.9tie
- WorldVQAtie36.835.9tie
- τ²-Bench Telecom (AA run)tie89.588.6tie
- MMMLUtie90.891.5tie
- vectara_answer_ratetie98.798tie
Questions people ask
Which is better, Claude Opus 4.5 or Claude Opus 4.7?
Claude Opus 4.7 wins all eight areas we test: coding, agents, reasoning, facts, images and charts, math, long documents and following instructions. Claude Opus 4.5 wins none.
Which is better for coding?
Claude Opus 4.7. It wins 8 of the 8 coding tests both models report; Claude Opus 4.5 wins none.
Which is cheaper?
Claude Opus 4.5 costs $5.00 per million input tokens and $25.00 per million output tokens; Claude Opus 4.7 costs $5.00 and $25.00. For a million tokens read plus a million written, they cost about the same.
How do you compare the two?
We use the 53 benchmark tests both models have published scores on. The verdict counts the 32 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 21 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.