Claude Sonnet 4.5 vs GPT-5 mini
Wins 4 of 8 areas
Coding · Agents · Reasoning · Facts
Wins 3 of 8 areas
Images and charts · Math · Following instructions
Claude Sonnet 4.5 wins more areas, narrowly.GPT-5 mini is cheaper and better at math.
Scores updated · 42 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 software40Claude Sonnet 4.54 of 4 tests
- FactsGetting facts right instead of making them up30Claude Sonnet 4.53 of 4 tests · 1 tie
- AgentsCarrying out multi-step tasks on its own20Claude Sonnet 4.52 of 3 tests · 1 tie
- ReasoningHard problems that need careful thinking21Claude Sonnet 4.52 of 4 tests · 1 tie
- MathCompetition and research-level math02GPT-5 mini2 of 2 tests
- Following instructionsDoing exactly what it is asked02GPT-5 mini2 of 2 tests
- Images and chartsUnderstanding pictures, charts and video01GPT-5 mini1 of 1 test
- Long documentsFinding answers in very long texts00Even0 each · 1 tie
Images and charts 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 mini costs 88% 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 4.5 pulls ahead
- Command-line tasks in a real terminalTerminal-Bench 2.1+52.1points ahead
- Fixes real GitHub issues in Python projectsSWE-bench Verified+17.4points ahead
- Abstract visual puzzles that people can solveARC-AGI-2+9.2points ahead
Where GPT-5 mini pulls ahead
- Unpublished advanced math problemsFrontierMath Tiers 1-3 (v2)+22.8points ahead
- Follows unfamiliar, precisely checkable instructionsIFBench+18.1points ahead
- Extremely hard research-level math problemsFrontierMath Tier 4+9.8points ahead
Every test, side by side
All 42 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingClaude Sonnet 4.5
- Terminal-Bench 2.1Claude Sonnet 4.5 by 52.155.83.7+52.1
- SWE-bench VerifiedClaude Sonnet 4.5 by 17.477.259.8+17.4
- SciCodeClaude Sonnet 4.5 by 6.745.739+6.7
- Terminal-Bench HardClaude Sonnet 4.5 by 2.335.633.3+2.3
AgentsClaude Sonnet 4.5
- τ-Bench V3 · BankingClaude Sonnet 4.5 by 924.515.5+9
- GDPValClaude Sonnet 4.5 by 720.513.5+7
- Terminal-Bench 4.0tie00tie
ReasoningClaude Sonnet 4.5
- ARC-AGI-2Claude Sonnet 4.5 by 9.213.64.4+9.2
- Humanity's Last ExamGPT-5 mini by 3.717.821.5+3.7
- CritPtClaude Sonnet 4.5 by 1.11.10+1.1
- GPQA Diamondtie83.482.8tie
FactsClaude Sonnet 4.5
- SimpleQA VerifiedClaude Sonnet 4.5 by 9.130.721.6+9.1
- AA-Omniscience · AccuracyClaude Sonnet 4.5 by 7.932.925+7.9
- AA-Omniscience · Non-hallucinationClaude Sonnet 4.5 by 7.350.943.6+7.3
- Vectara HHEM hallucination ratelower is bettertie1212.9tie
MathGPT-5 mini
- FrontierMath Tiers 1-3 (v2)GPT-5 mini by 22.823.946.7+22.8
- FrontierMath Tier 4GPT-5 mini by 9.82.412.2+9.8
Following instructionsGPT-5 mini
- IFBenchGPT-5 mini by 18.157.375.4+18.1
- Multi-ChallengeGPT-5 mini by 3.755.359+3.7
Other results21 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.
- Artificial Analysis Coding IndexClaude Sonnet 4.5 by 36.552.115.6+36.5
- AA-OmniscienceClaude Sonnet 4.5 by 17.2-0.1-17.3+17.2
- HMMT 2025GPT-5 mini by 13.274.687.8+13.2
- swe_bench_bashClaude Sonnet 4.5 by 11.671.459.8+11.6
- τ²-Bench Telecom (AA run)Claude Sonnet 4.5 by 9.778.168.4+9.7
- ARC-AGI-1Claude Sonnet 4.5 by 9.463.754.3+9.4
- AA Agentic IndexClaude Sonnet 4.5 by 8.617.58.9+8.6
- HarmBenchGPT-5 mini by 5.19297.1+5.1
- vectara_answer_rateGPT-5 mini by 4.395.699.9+4.3
- vectara_avg_summary_lengthGPT-5 mini by 41.9 rating points127.8169.7+41.9 rating
- AIME 2025GPT-5 mini by 4.18791.1+4.1
- AIR-Bench 2024Claude Sonnet 4.5 by 4.189.885.7+4.1
- AA IntelligenceClaude Sonnet 4.5 by 3.920.716.8+3.9
- bbqClaude Sonnet 4.5 by 2.698.996.3+2.6
- HMMT Nov. 2025GPT-5 mini by 2.581.784.2+2.5
- matharena_visual_math_overallGPT-5 mini by 2.475.878.2+2.4
- frontiermath_tier_4_v1GPT-5 mini by 2.14.26.3+2.1
- XSTestGPT-5 mini by 1.396.497.7+1.3
- anthropic_red_teamtie98.299.1tie
- vectara_factual_consistencytie8887.1tie
- simple_safety_teststie100100tie
Questions people ask
Which is better, Claude Sonnet 4.5 or GPT-5 mini?
Claude Sonnet 4.5 wins four of the eight areas we test: coding, agents, reasoning and facts. GPT-5 mini wins images and charts, math and following instructions, and costs 88% less. They are level on long documents.
Which is better for coding?
Claude Sonnet 4.5. It wins 4 of the 4 coding tests both models report; GPT-5 mini wins none.
Which is cheaper?
Claude Sonnet 4.5 costs $3.00 per million input tokens and $15.00 per million output tokens; GPT-5 mini costs $0.25 and $2.00. That makes GPT-5 mini about 88% cheaper for the same work.
How do you compare the two?
We use the 42 benchmark tests both models have published scores on. The verdict counts the 21 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 21 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.