Claude Opus 4.5 vs GPT-5
Wins 3 of 8 areas
Coding · Reasoning · Facts
Wins 3 of 8 areas
Images and charts · Math · Following instructions
The two are evenly matched.Claude Opus 4.5 is better at coding and reasoning; GPT-5 at images and charts and following instructions.
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.
- CodingWriting and fixing software52Claude Opus 4.55 of 7 tests
- ReasoningHard problems that need careful thinking31Claude Opus 4.53 of 4 tests
- FactsGetting facts right instead of making them up31Claude Opus 4.53 of 4 tests
- Images and chartsUnderstanding pictures, charts and video02GPT-52 of 3 tests · 1 tie
- Following instructionsDoing exactly what it is asked02GPT-52 of 2 tests
- MathCompetition and research-level math12GPT-52 of 3 tests
- AgentsCarrying out multi-step tasks on its own11Even1 each
- Long documentsFinding answers in very long texts00Even0 each · 1 tie
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.
GPT-5 costs 63% 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 4.5 pulls ahead
- Abstract visual puzzles that people can solveARC-AGI-2+27.7points ahead
- Real work tasks from 44 professionsGDPVal+26.3points ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+21.2points ahead
Where GPT-5 pulls ahead
- Fixes real GitHub issues in many programming languagesSWE-bench Multilingual+48.3points ahead
- Unpublished advanced math problemsFrontierMath Tiers 1-3 (v2)+21points ahead
- Finds hard-to-locate facts by browsing the webBrowseComp+17.9points 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 Opus 4.5
- SWE-bench MultilingualGPT-5 by 48.3755.3+48.3
- Terminal-Bench HardClaude Opus 4.5 by 14.44732.6+14.4
- SWE-bench ProClaude Opus 4.5 by 10.25241.8+10.2
- SciCodeClaude Opus 4.5 by 6.649.542.9+6.6
- SWE-bench VerifiedClaude Opus 4.5 by 680.974.9+6
- LMArena · WebDevClaude Opus 4.5 by 51 rating points14691418+51 rating
- LiveCodeBench v6GPT-5 by 2.284.887+2.2
AgentsEven
- GDPValClaude Opus 4.5 by 26.347.321+26.3
- BrowseCompGPT-5 by 17.93754.9+17.9
ReasoningClaude Opus 4.5
- ARC-AGI-2Claude Opus 4.5 by 27.737.69.9+27.7
- Humanity's Last ExamClaude Opus 4.5 by 1.630.128.5+1.6
- GPQA DiamondClaude Opus 4.5 by 1.286.685.4+1.2
- CritPtGPT-5 by 1.14.65.7+1.1
FactsClaude Opus 4.5
- AA-Omniscience · Non-hallucinationClaude Opus 4.5 by 21.23917.8+21.2
- AA-Omniscience · AccuracyClaude Opus 4.5 by 6.346.640.3+6.3
- SimpleQA VerifiedGPT-5 by 4.445.750.1+4.4
- Vectara HHEM hallucination ratelower is betterClaude Opus 4.5 by 4.210.915.1+4.2
Images and chartsGPT-5
- CharXiv (RQ)GPT-5 by 13.967.281.1+13.9
- MMMUGPT-5 by 3.580.784.2+3.5
- MMMU-Protie7474.2tie
MathGPT-5
- FrontierMath Tiers 1-3 (v2)GPT-5 by 2134.455.4+21
- FrontierMath Tier 4GPT-5 by 17.14.921.9+17.1
- IMOAnswerBenchClaude Opus 4.5 by 88476+8
Following instructionsGPT-5
- IFBenchGPT-5 by 15.15873.1+15.1
- Multi-ChallengeGPT-5 by 10.65969.6+10.6
Other results26 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 33.159.626.5+33.1
- Terminal-Bench 2.0Claude Opus 4.5 by 24.159.335.2+24.1
- AA-OmniscienceClaude Opus 4.5 by 22.714-8.7+22.7
- ARC-AGI-1Claude Opus 4.5 by 16.38265.7+16.3
- τ²-BenchClaude Opus 4.5 by 15.898.282.4+15.8
- Artificial Analysis Coding IndexClaude Opus 4.5 by 1047.837.8+10
- swe_bench_bashClaude Opus 4.5 by 9.474.465+9.4
- frontiermath_tier_4_v1GPT-5 by 8.34.212.5+8.3
- HLE (with tools)Claude Opus 4.5 by 843.235.2+8
- τ²-Bench (Retail)Claude Opus 4.5 by 7.888.981.1+7.8
- AA IntelligenceClaude Opus 4.5 by 6.129.123+6.1
- metr_hcastClaude Opus 4.5 by 573.468.4+5
- vectara_avg_summary_lengthGPT-5 by 48.2 rating points114.5162.7+48.2 rating
- τ²-Bench Telecom (AA run)Claude Opus 4.5 by 4.789.584.8+4.7
- HMMT Feb. 2025Claude Opus 4.5 by 4.692.988.3+4.6
- vectara_factual_consistencyClaude Opus 4.5 by 4.289.184.9+4.2
- HMMT Nov. 2025Claude Opus 4.5 by 4.193.389.2+4.1
- AIME25 no toolsGPT-5 by 3.69194.6+3.6
- LiveCodeBenchClaude Opus 4.5 by 3.58783.5+3.5
- MMLU-ProClaude Opus 4.5 by 2.489.587.1+2.4
- AIME 2025GPT-5 by 1.892.894.6+1.8
- OJ-Bench (cpp)GPT-5 by 1.654.656.2+1.6
- vectara_answer_rateGPT-5 by 1.298.799.9+1.2
- MMLU-Reduxtie95.695.3tie
- VideoMMMUtie84.484.6tie
- metr_re_benchtie4040tie
Questions people ask
Which is better, Claude Opus 4.5 or GPT-5?
Claude Opus 4.5 and GPT-5 each win three of the eight areas we test. Claude Opus 4.5 wins coding, reasoning and facts; GPT-5 wins images and charts, math and following instructions. GPT-5 costs 63% less. They are level on agents and long documents.
Which is better for coding?
Claude Opus 4.5. It wins 5 of the 7 coding tests both models report; GPT-5 wins 2.
Which is cheaper?
Claude Opus 4.5 costs $5.00 per million input tokens and $25.00 per million output tokens; GPT-5 costs $1.25 and $10.00. That makes GPT-5 about 63% 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 26 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 26 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.