Claude Sonnet 4.6 vs GPT-5.4 mini
Wins 7 of 8 areas
Coding · Agents · Reasoning · Facts · Images and charts · Math · Long documents
Wins 0 of 8 areas
—
Claude Sonnet 4.6 is the stronger all-rounder.GPT-5.4 mini is cheaper.
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.
- AgentsCarrying out multi-step tasks on its own50Claude Sonnet 4.65 of 7 tests · 2 ties
- CodingWriting and fixing software41Claude Sonnet 4.64 of 7 tests · 2 ties
- ReasoningHard problems that need careful thinking31Claude Sonnet 4.63 of 5 tests · 1 tie
- FactsGetting facts right instead of making them up31Claude Sonnet 4.63 of 4 tests
- Images and chartsUnderstanding pictures, charts and video10Claude Sonnet 4.61 of 2 tests · 1 tie
- MathCompetition and research-level math10Claude Sonnet 4.61 of 1 test
- Long documentsFinding answers in very long texts10Claude Sonnet 4.61 of 1 test
- Following instructionsDoing exactly what it is asked11Even1 each
Math and long documents 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.
GPT-5.4 mini costs 71% 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 Sonnet 4.6 pulls ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+41.8points ahead
- Abstract visual puzzles that people can solveARC-AGI-2+39.4points ahead
- Fresh logic puzzles that change regularlyLiveBench · Reasoning+13.5points 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 Sonnet 4.6
- LMArena · WebDevClaude Sonnet 4.6 by 125 rating points15221397+125 rating
- Terminal-Bench 2.1Claude Sonnet 4.6 by 1271.259.2+12
- LiveBench · CodingClaude Sonnet 4.6 by 7.779.371.6+7.7
- SWE-bench ProClaude Sonnet 4.6 by 3.758.154.4+3.7
- SciCodeGPT-5.4 mini by 250.152.1+2
- LiveBench · Agentic Codingtie42.641.7tie
- Terminal-Bench Hardtie5352.3tie
AgentsClaude Sonnet 4.6
- GDPValClaude Sonnet 4.6 by 10.936.725.8+10.9
- τ-Bench V3 · BankingClaude Sonnet 4.6 by 8.834.425.6+8.8
- AA IT-Bench SREClaude Sonnet 4.6 by 4.639.835.2+4.6
- MCP AtlasClaude Sonnet 4.6 by 3.661.357.7+3.6
- Terminal-Bench 4.0Claude Sonnet 4.6 by 132+1
- OSWorld-Verifiedtie72.572.1tie
- AA ApexAgentstie2828.2tie
ReasoningClaude Sonnet 4.6
- ARC-AGI-2Claude Sonnet 4.6 by 39.458.318.9+39.4
- LiveBench · ReasoningClaude Sonnet 4.6 by 13.584.871.3+13.5
- CritPtGPT-5.4 mini by 6.93.110+6.9
- Humanity's Last ExamClaude Sonnet 4.6 by 5.533.628.1+5.5
- GPQA Diamondtie87.587.5tie
FactsClaude Sonnet 4.6
- AA-Omniscience · Non-hallucinationClaude Sonnet 4.6 by 41.851.69.8+41.8
- Vectara HHEM hallucination ratelower is betterGPT-5.4 mini by 5.110.65.5+5.1
- AA-Omniscience · AccuracyClaude Sonnet 4.6 by 3.440.937.5+3.4
- SimpleQA VerifiedClaude Sonnet 4.6 by 3.432.829.4+3.4
Images and chartsClaude Sonnet 4.6
- LMArena · VisionClaude Sonnet 4.6 by 38 rating points12831245+38 rating
- MMMU-Protie73.373.3tie
Following instructionsEven
- IFBenchGPT-5.4 mini by 16.756.673.3+16.7
- LiveBench · Instruction FollowingClaude Sonnet 4.6 by 3.463.259.8+3.4
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 Sonnet 4.6 by 31.112.2-18.9+31.1
- ARC-AGI-1Claude Sonnet 4.6 by 22.38663.7+22.3
- AA Agentic IndexClaude Sonnet 4.6 by 13.433.119.7+13.4
- τ²-Bench Telecom (AA run)GPT-5.4 mini by 7.675.783.3+7.6
- livebench_data_analysisClaude Sonnet 4.6 by 7.27870.8+7.2
- Artificial Analysis Coding IndexClaude Sonnet 4.6 by 6.96356.1+6.9
- frontiermath_tier_4_v1Claude Sonnet 4.6 by 6.28.32.1+6.2
- AA IntelligenceClaude Sonnet 4.6 by 630.124.1+6
- vectara_avg_summary_lengthClaude Sonnet 4.6 by 60 rating points114.754.7+60 rating
- Finance Agent v2Claude Sonnet 4.6 by 5.65145.4+5.6
- Legal Agent BenchmarkClaude Sonnet 4.6 by 5.45.40+5.4
- livebench_languageClaude Sonnet 4.6 by 5.176.171+5.1
- vectara_factual_consistencyGPT-5.4 mini by 5.189.494.5+5.1
- ToolathlonGPT-5.4 mini by 1.94142.9+1.9
- Terminal-Bench 2.0tie59.160tie
- vectara_answer_ratetie99.9100tie
Questions people ask
Which is better, Claude Sonnet 4.6 or GPT-5.4 mini?
Claude Sonnet 4.6 wins seven of the eight areas we test: coding, agents, reasoning, facts, images and charts, math and long documents. GPT-5.4 mini wins none, but costs 71% less. They are level on following instructions.
Which is better for coding?
Claude Sonnet 4.6. It wins 4 of the 7 coding tests both models report; GPT-5.4 mini wins 1, and 2 are ties.
Which is cheaper?
Claude Sonnet 4.6 costs $3.00 per million input tokens and $15.00 per million output tokens; GPT-5.4 mini costs $0.75 and $4.50. That makes GPT-5.4 mini about 71% 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.