Claude Opus 4.7 vs GPT-5.5
Wins 0 of 8 areas
—
Wins 8 of 8 areas
Coding · Agents · Reasoning · Facts · Images and charts · Math · Long documents · Following instructions
GPT-5.5 is the stronger all-rounder.Claude Opus 4.7 is cheaper.
Both rank among the ten best models we track in coding.
Scores updated · 82 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 thinking06GPT-5.56 of 7 tests · 1 tie
- MathCompetition and research-level math06GPT-5.56 of 6 tests
- CodingWriting and fixing software24GPT-5.54 of 7 tests · 1 tie
- AgentsCarrying out multi-step tasks on its own13GPT-5.53 of 7 tests · 3 ties
- Images and chartsUnderstanding pictures, charts and video13GPT-5.53 of 5 tests · 1 tie
- FactsGetting facts right instead of making them up13GPT-5.53 of 4 tests
- Following instructionsDoing exactly what it is asked02GPT-5.52 of 2 tests
- Long documentsFinding answers in very long texts01GPT-5.51 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.
Claude Opus 4.7 costs 14% less for the same work.
The biggest differences
The three tests each model wins by the widest margin. Scores are out of 100.
Where GPT-5.5 pulls ahead
- Extremely hard research-level math problemsFrontierMath Tier 4+40.8points ahead
- Proof problems from the US Math OlympiadUSAMO 2026+28.9points ahead
- Follows unfamiliar, precisely checkable instructionsIFBench+17.3points ahead
Where Claude Opus 4.7 pulls ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+46.7points ahead
- Long, multi-file coding tasks in real codebasesSWE-bench Pro+5.7points ahead
- Multi-step tasks using many real software toolsMCP Atlas+2points ahead
Every test, side by side
All 82 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingGPT-5.5
- Terminal-Bench HardGPT-5.5 by 9.151.560.6+9.1
- SWE-bench ProClaude Opus 4.7 by 5.764.358.6+5.7
- LMArena · WebDevClaude Opus 4.7 by 45 rating points15581513+45 rating
- LiveBench · Agentic CodingGPT-5.5 by 3.350.754+3.3
- SciCodeGPT-5.5 by 1.354.555.8+1.3
- Terminal-Bench 2.1GPT-5.5 by 1.283.184.3+1.2
- LiveBench · Codingtie82.182.2tie
AgentsGPT-5.5
- BrowseCompGPT-5.5 by 5.179.384.4+5.1
- τ-Bench V3 · BankingGPT-5.5 by 4.434.639+4.4
- AA ApexAgentsGPT-5.5 by 3.833.937.7+3.8
- MCP AtlasClaude Opus 4.7 by 277.375.3+2
- AA IT-Bench SREtie46.745.8tie
- OSWorld-Verifiedtie7878.7tie
- GDPValtie42.842.7tie
ReasoningGPT-5.5
- CritPtGPT-5.5 by 15.11227.1+15.1
- ARC-AGI-2GPT-5.5 by 9.275.885+9.2
- SimpleBenchGPT-5.5 by 7.361.769+7.3
- Humanity's Last ExamGPT-5.5 by 3.542.345.8+3.5
- LiveBench · ReasoningGPT-5.5 by 2.587.289.7+2.5
- GPQA DiamondGPT-5.5 by 2.191.493.5+2.1
- ARC-AGI-3tie0.20.4tie
FactsGPT-5.5
- AA-Omniscience · Non-hallucinationClaude Opus 4.7 by 46.757.711+46.7
- SimpleQA VerifiedGPT-5.5 by 11.351.763+11.3
- AA-Omniscience · AccuracyGPT-5.5 by 9.148.958+9.1
- Vectara HHEM hallucination ratelower is betterGPT-5.5 by 2.7129.3+2.7
Images and chartsGPT-5.5
- BLINKGPT-5.5 by 7.970.478.3+7.9
- OCRBenchv2GPT-5.5 by 4.256.961.1+4.2
- LMArena · VisionClaude Opus 4.7 by 21 rating points13131292+21 rating
- MMMU-ProGPT-5.5 by 1.178.879.9+1.1
- MathVistatie84.484.2tie
MathGPT-5.5
- FrontierMath Tier 4GPT-5.5 by 40.831.772.5+40.8
- USAMO 2026GPT-5.5 by 28.969.398.2+28.9
- FrontierMath Tiers 1-3 (v2)GPT-5.5 by 15.170.285.3+15.1
- HMMT Feb. 2026GPT-5.5 by 4.693.998.5+4.6
- AIME 2026GPT-5.5 by 4.295.8100+4.2
- LiveBench · MathematicsGPT-5.5 by 392.895.9+3
Following instructionsGPT-5.5
- IFBenchGPT-5.5 by 17.358.675.9+17.3
- LiveBench · Instruction FollowingGPT-5.5 by 466.770.7+4
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.
- BabyVisionGPT-5.5 by 33.722.255.9+33.7
- OfficeQA ProClaude Opus 4.7 by 26.580.654.1+26.5
- MMSIBench (circular)GPT-5.5 by 18.617.436+18.6
- DeepSWEGPT-5.5 by 165470+16
- Terminal-Bench 2.0GPT-5.5 by 13.369.482.7+13.3
- DynaMathGPT-5.5 by 12.863.175.9+12.8
- frontiermath_tier_4_v1GPT-5.5 by 12.522.935.4+12.5
- ERQAGPT-5.5 by 1252.564.5+12
- VisuLogicGPT-5.5 by 10.932.643.5+10.9
- livebench_languageGPT-5.5 by 9.577.987.4+9.5
- CyberGymGPT-5.5 by 8.773.181.8+8.7
- MathVerse (Vision-Only)GPT-5.5 by 7.277.484.6+7.2
- Office QA Pro [Multimodal]Claude Opus 4.7 by 776.569.5+7
- AA-OmniscienceClaude Opus 4.7 by 6.827.320.5+6.8
- RealWorldQAGPT-5.5 by 6.675.682.2+6.6
- SWE-Pro BenchClaude Opus 4.7 by 5.764.358.6+5.7
- τ²-Bench Telecom (AA run)GPT-5.5 by 5.388.693.9+5.3
- Legal Agent BenchmarkClaude Opus 4.7 by 57.12.1+5
- ZeroBench (main)GPT-5.5 by 5813+5
- EmbSpatial-BenchGPT-5.5 by 4.777.281.9+4.7
- Finance AgentClaude Opus 4.7 by 4.464.460+4.4
- ChartQAProGPT-5.5 by 3.965.569.4+3.9
- ZeroBench (sub)GPT-5.5 by 3.937.141+3.9
- LiveBenchGPT-5.5 by 3.876.980.7+3.8
- ToolathlonClaude Opus 4.7 by 3.759.355.6+3.7
- GraphWalks — Parents task, 256K-token context subsetClaude Opus 4.7 by 3.593.690.1+3.5
- livebench_data_analysisGPT-5.5 by 3.378.381.6+3.3
- GraphWalks BFS 256KClaude Opus 4.7 by 3.276.973.7+3.2
- ARC-AGI-1GPT-5.5 by 39295+3
- DUDEClaude Opus 4.7 by 2.884.581.7+2.8
- vectara_factual_consistencyGPT-5.5 by 2.78890.7+2.7
- HLE (with tools)Claude Opus 4.7 by 2.554.752.2+2.5
- AA IntelligenceClaude Opus 4.7 by 2.340.738.4+2.3
- AA Agentic IndexClaude Opus 4.7 by 2.239.537.3+2.2
- SimpleVQAGPT-5.5 by 2.156.558.6+2.1
- HiL-Bench (Tools-allowed)Claude Opus 4.7 by 241.739.7+2
- vectara_answer_rateGPT-5.5 by 298100+2
- vectara_avg_summary_lengthClaude Opus 4.7 by 19.5 rating points149.1129.6+19.5 rating
- GDPval-AA (Elo)GPT-5.5 by 16 rating points17531769+16 rating
- Artificial Analysis Coding IndexGPT-5.5 by 1.373.674.9+1.3
- WorldVQAClaude Opus 4.7 by 1.335.934.6+1.3
- Finance Agent v1.1tie64.465.3tie
- Finance Agent v2tie51.551.8tie
Questions people ask
Which is better, Claude Opus 4.7 or GPT-5.5?
GPT-5.5 wins all eight areas we test: coding, agents, reasoning, facts, images and charts, math, long documents and following instructions. Claude Opus 4.7 wins none, but costs 14% less.
Which is better for coding?
GPT-5.5. It wins 4 of the 7 coding tests both models report; Claude Opus 4.7 wins 2, and 1 is a tie.
Which is cheaper?
Claude Opus 4.7 costs $5.00 per million input tokens and $25.00 per million output tokens; GPT-5.5 costs $5.00 and $30.00. That makes Claude Opus 4.7 about 14% cheaper for the same work.
How do you compare the two?
We use the 82 benchmark tests both models have published scores on. The verdict counts the 39 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.