Claude Opus 4.7 vs GPT-5.4 nano
Wins 7 of 8 areas
Coding · Agents · Reasoning · Facts · Images and charts · Math · Long documents
Wins 1 of 8 areas
Following instructions
Claude Opus 4.7 is the stronger all-rounder.GPT-5.4 nano is cheaper and better at following instructions.
Scores updated · 45 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 software60Claude Opus 4.76 of 6 tests
- AgentsCarrying out multi-step tasks on its own60Claude Opus 4.76 of 6 tests
- ReasoningHard problems that need careful thinking50Claude Opus 4.75 of 5 tests
- MathCompetition and research-level math30Claude Opus 4.73 of 3 tests
- FactsGetting facts right instead of making them up31Claude Opus 4.73 of 4 tests
- Images and chartsUnderstanding pictures, charts and video20Claude Opus 4.72 of 2 tests
- Long documentsFinding answers in very long texts10Claude Opus 4.71 of 1 test
- Following instructionsDoing exactly what it is asked01GPT-5.4 nano1 of 2 tests · 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-5.4 nano costs 95% 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 4.7 pulls ahead
- Abstract visual puzzles that people can solveARC-AGI-2+70.1points ahead
- Short factual questions, answered correctlySimpleQA Verified+40points ahead
- Completes tasks by operating a computer desktopOSWorld-Verified+39points ahead
Where GPT-5.4 nano pulls ahead
- Follows unfamiliar, precisely checkable instructionsIFBench+17.3points ahead
Every test, side by side
All 45 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingClaude Opus 4.7
- Terminal-Bench 2.1Claude Opus 4.7 by 22.483.160.7+22.4
- SWE-bench ProClaude Opus 4.7 by 11.964.352.4+11.9
- LiveBench · CodingClaude Opus 4.7 by 11.382.170.8+11.3
- Terminal-Bench HardClaude Opus 4.7 by 9.151.542.4+9.1
- SciCodeClaude Opus 4.7 by 7.354.547.2+7.3
- LiveBench · Agentic CodingClaude Opus 4.7 by 3.950.746.8+3.9
AgentsClaude Opus 4.7
- OSWorld-VerifiedClaude Opus 4.7 by 397839+39
- AA IT-Bench SREClaude Opus 4.7 by 22.346.724.4+22.3
- MCP AtlasClaude Opus 4.7 by 21.277.356.1+21.2
- GDPValClaude Opus 4.7 by 20.342.822.5+20.3
- AA ApexAgentsClaude Opus 4.7 by 933.924.9+9
- τ-Bench V3 · BankingClaude Opus 4.7 by 7.234.627.4+7.2
ReasoningClaude Opus 4.7
- ARC-AGI-2Claude Opus 4.7 by 70.175.85.7+70.1
- Humanity's Last ExamClaude Opus 4.7 by 1442.328.3+14
- GPQA DiamondClaude Opus 4.7 by 9.791.481.7+9.7
- LiveBench · ReasoningClaude Opus 4.7 by 6.187.281.1+6.1
- CritPtClaude Opus 4.7 by 2.7129.3+2.7
FactsClaude Opus 4.7
- SimpleQA VerifiedClaude Opus 4.7 by 4051.711.7+40
- AA-Omniscience · Non-hallucinationClaude Opus 4.7 by 31.957.725.8+31.9
- AA-Omniscience · AccuracyClaude Opus 4.7 by 23.248.925.7+23.2
- Vectara HHEM hallucination ratelower is betterGPT-5.4 nano by 8.9123.1+8.9
Images and chartsClaude Opus 4.7
- MMMU-ProClaude Opus 4.7 by 13.478.865.4+13.4
- LMArena · VisionClaude Opus 4.7 by 117 rating points13131196+117 rating
MathClaude Opus 4.7
- FrontierMath Tiers 1-3 (v2)Claude Opus 4.7 by 25.370.244.9+25.3
- FrontierMath Tier 4Claude Opus 4.7 by 19.531.712.2+19.5
- LiveBench · MathematicsClaude Opus 4.7 by 1.992.891+1.9
Following instructionsGPT-5.4 nano
- IFBenchGPT-5.4 nano by 17.358.675.9+17.3
- LiveBench · Instruction Followingtie66.767.2tie
Other results16 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-OmniscienceClaude Opus 4.7 by 56.827.3-29.5+56.8
- ARC-AGI-1Claude Opus 4.7 by 40.59251.5+40.5
- ToolathlonClaude Opus 4.7 by 23.859.335.5+23.8
- Terminal-Bench 2.0Claude Opus 4.7 by 23.169.446.3+23.1
- AA Agentic IndexClaude Opus 4.7 by 21.839.517.7+21.8
- AA IntelligenceClaude Opus 4.7 by 2040.720.7+20
- Artificial Analysis Coding IndexClaude Opus 4.7 by 17.573.656.1+17.5
- frontiermath_tier_4_v1Claude Opus 4.7 by 16.622.96.3+16.6
- livebench_languageClaude Opus 4.7 by 15.477.962.5+15.4
- Finance Agent v2Claude Opus 4.7 by 13.351.538.2+13.3
- τ²-Bench Telecom (AA run)Claude Opus 4.7 by 12.688.676+12.6
- livebench_data_analysisClaude Opus 4.7 by 10.778.367.6+10.7
- vectara_factual_consistencyGPT-5.4 nano by 8.98896.9+8.9
- LiveBenchClaude Opus 4.7 by 6.876.970.1+6.8
- vectara_answer_rateGPT-5.4 nano by 298100+2
- vectara_avg_summary_lengthtie149.1144.4tie
Questions people ask
Which is better, Claude Opus 4.7 or GPT-5.4 nano?
Claude Opus 4.7 wins seven of the eight areas we test: coding, agents, reasoning, facts, images and charts, math and long documents. GPT-5.4 nano wins following instructions, and costs 95% less.
Which is better for coding?
Claude Opus 4.7. It wins 6 of the 6 coding tests both models report; GPT-5.4 nano wins none.
Which is cheaper?
Claude Opus 4.7 costs $5.00 per million input tokens and $25.00 per million output tokens; GPT-5.4 nano costs $0.20 and $1.25. That makes GPT-5.4 nano about 95% cheaper for the same work.
How do you compare the two?
We use the 45 benchmark tests both models have published scores on. The verdict counts the 29 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 16 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.