Claude Sonnet 5 vs GPT-5.4
Wins 2 of 8 areas
Coding · Agents
Wins 5 of 8 areas
Reasoning · Facts · Images and charts · Math · Following instructions
GPT-5.4 is the stronger all-rounder.Claude Sonnet 5 is cheaper and better at coding.
Scores updated · 40 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.
- MathCompetition and research-level math04GPT-5.44 of 4 tests
- ReasoningHard problems that need careful thinking02GPT-5.42 of 4 tests · 2 ties
- FactsGetting facts right instead of making them up12GPT-5.42 of 3 tests
- Images and chartsUnderstanding pictures, charts and video12GPT-5.42 of 3 tests
- Following instructionsDoing exactly what it is asked01GPT-5.41 of 1 test
- CodingWriting and fixing software61Claude Sonnet 56 of 7 tests
- AgentsCarrying out multi-step tasks on its own31Claude Sonnet 53 of 4 tests
- Long documentsFinding answers in very long texts00Even0 each · 1 tie
Long documents and following instructions rest on a single test each.
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 5 costs 31% 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-5.4 pulls ahead
- Extremely hard research-level math problemsFrontierMath Tier 4+19.7points ahead
- Proof problems from the US Math OlympiadUSAMO 2026+15.7points ahead
- Unpublished advanced math problemsFrontierMath Tiers 1-3 (v2)+13points ahead
Where Claude Sonnet 5 pulls ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+52.3points ahead
- Real work tasks from 44 professionsGDPVal+10.9points ahead
- Fixes real GitHub issues in many programming languagesSWE-bench Multilingual+6.6points ahead
Every test, side by side
All 40 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingClaude Sonnet 5
- LMArena · WebDevClaude Sonnet 5 by 76 rating points15411465+76 rating
- SWE-bench MultilingualClaude Sonnet 5 by 6.678.371.7+6.6
- LiveBench · Agentic CodingClaude Sonnet 5 by 5.659.453.8+5.6
- SWE-bench ProClaude Sonnet 5 by 5.563.257.7+5.5
- LiveBench · CodingClaude Sonnet 5 by 3.280.777.5+3.2
- SciCodeGPT-5.4 by 2.354.356.6+2.3
- Terminal-Bench 2.1Claude Sonnet 5 by 2.280.578.3+2.2
AgentsClaude Sonnet 5
- GDPValClaude Sonnet 5 by 10.948.337.4+10.9
- OSWorld-VerifiedClaude Sonnet 5 by 6.281.275+6.2
- τ-Bench V3 · BankingGPT-5.4 by 2.337.339.6+2.3
- BrowseCompClaude Sonnet 5 by 284.782.7+2
ReasoningGPT-5.4
- CritPtGPT-5.4 by 6.516.923.4+6.5
- Humanity's Last ExamGPT-5.4 by 2.441.343.7+2.4
- GPQA Diamondtie91.192tie
- LiveBench · Reasoningtie88.788.1tie
FactsGPT-5.4
- AA-Omniscience · Non-hallucinationClaude Sonnet 5 by 52.360.68.3+52.3
- SimpleQA VerifiedGPT-5.4 by 11.433.745.1+11.4
- AA-Omniscience · AccuracyGPT-5.4 by 10.84050.9+10.8
Images and chartsGPT-5.4
- CharXiv (RQ)Claude Sonnet 5 by 5.588.382.8+5.5
- LMArena · VisionGPT-5.4 by 27 rating points12751302+27 rating
- MMMU-ProGPT-5.4 by 1.177.378.4+1.1
MathGPT-5.4
- FrontierMath Tier 4GPT-5.4 by 19.729.349+19.7
- USAMO 2026GPT-5.4 by 15.779.595.2+15.7
- FrontierMath Tiers 1-3 (v2)GPT-5.4 by 1365.678.6+13
- LiveBench · MathematicsGPT-5.4 by 1.392.994.2+1.3
Following instructionsGPT-5.4
- LiveBench · Instruction FollowingGPT-5.4 by 6.363.970.2+6.3
Other results13 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 Sonnet 5 by 10.716.45.8+10.7
- OfficeQA ProClaude Sonnet 5 by 8.359.451.1+8.3
- livebench_data_analysisGPT-5.4 by 7.671.779.3+7.6
- livebench_languageGPT-5.4 by 7.67582.6+7.6
- OfficeQAClaude Sonnet 5 by 775.168.1+7
- HLE (with tools)Claude Sonnet 5 by 5.457.452+5.4
- Legal Agent BenchmarkClaude Sonnet 5 by 5.45.80.4+5.4
- Terminal-Bench 2.0Claude Sonnet 5 by 5.380.475.1+5.3
- DeepSWE 1.1Claude Sonnet 5 by 25452+2
- AA Intelligencetie38.239tie
- Artificial Analysis Coding Indextie71.571.1tie
- Toolathlontie54.354.6tie
- AA Agentic Indextie44.344.2tie
Questions people ask
Which is better, Claude Sonnet 5 or GPT-5.4?
GPT-5.4 wins five of the eight areas we test: reasoning, facts, images and charts, math and following instructions. Claude Sonnet 5 wins coding and agents, and costs 31% less. They are level on long documents.
Which is better for coding?
Claude Sonnet 5. It wins 6 of the 7 coding tests both models report; GPT-5.4 wins 1.
Which is cheaper?
Claude Sonnet 5 costs $2.00 per million input tokens and $10.00 per million output tokens; GPT-5.4 costs $2.50 and $15.00. That makes Claude Sonnet 5 about 31% cheaper for the same work.
How do you compare the two?
We use the 40 benchmark tests both models have published scores on. The verdict counts the 27 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 13 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.