GPT-5 vs o4-mini
Wins 7 of 8 areas
Coding · Reasoning · Facts · Images and charts · Math · Long documents · Following instructions
Wins 0 of 8 areas
—
GPT-5 is the stronger all-rounder.o4-mini is cheaper.
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.
- ReasoningHard problems that need careful thinking40GPT-54 of 4 tests
- Images and chartsUnderstanding pictures, charts and video40GPT-54 of 4 tests
- FactsGetting facts right instead of making them up31GPT-53 of 4 tests
- MathCompetition and research-level math20GPT-52 of 2 tests
- Following instructionsDoing exactly what it is asked20GPT-52 of 2 tests
- CodingWriting and fixing software21GPT-52 of 3 tests
- Long documentsFinding answers in very long texts10GPT-51 of 1 test
- AgentsCarrying out multi-step tasks on its own11Even1 each
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.
o4-mini costs 51% 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 pulls ahead
- Short factual questions, answered correctlySimpleQA Verified+31.3points ahead
- Keeps track of context across a multi-turn chatMulti-Challenge+26.6points ahead
- Unpublished advanced math problemsFrontierMath Tiers 1-3 (v2)+19.3points ahead
Where o4-mini pulls ahead
- Real work tasks from 44 professionsGDPVal+4.4points ahead
- Code for real scientific research problemsSciCode+3.6points ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+1.7points 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.
CodingGPT-5
- Terminal-Bench HardGPT-5 by 17.432.615.2+17.4
- SWE-bench VerifiedGPT-5 by 6.874.968.1+6.8
- SciCodeo4-mini by 3.642.946.5+3.6
ReasoningGPT-5
- Humanity's Last ExamGPT-5 by 1228.516.5+12
- GPQA DiamondGPT-5 by 785.478.4+7
- CritPtGPT-5 by 5.15.70.6+5.1
- ARC-AGI-2GPT-5 by 3.89.96.1+3.8
FactsGPT-5
- SimpleQA VerifiedGPT-5 by 31.350.118.8+31.3
- AA-Omniscience · AccuracyGPT-5 by 15.540.324.8+15.5
- Vectara HHEM hallucination ratelower is betterGPT-5 by 3.515.118.6+3.5
- AA-Omniscience · Non-hallucinationo4-mini by 1.717.819.5+1.7
Images and chartsGPT-5
- CharXiv (RQ)GPT-5 by 9.181.172+9.1
- MMMU-ProGPT-5 by 574.269.2+5
- MMMUGPT-5 by 2.684.281.6+2.6
- LMArena · VisionGPT-5 by 14 rating points12081194+14 rating
MathGPT-5
- FrontierMath Tiers 1-3 (v2)GPT-5 by 19.355.436.1+19.3
- FrontierMath Tier 4GPT-5 by 17.121.94.9+17.1
Following instructionsGPT-5
- Multi-ChallengeGPT-5 by 26.669.643+26.6
- IFBenchGPT-5 by 4.473.168.7+4.4
Other results20 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.
- τ²-Bench Telecom (AA run)GPT-5 by 29.284.855.6+29.2
- AA-OmniscienceGPT-5 by 27-8.7-35.7+27
- imo_2025GPT-5 by 23.838.114.3+23.8
- Aider-PolyglotGPT-5 by 19.18868.9+19.1
- Artificial Analysis Coding IndexGPT-5 by 12.237.825.6+12.2
- AA Agentic Indexo4-mini by 9.626.536.1+9.6
- AIR-Bench 2024GPT-5 by 9.287.778.5+9.2
- ARC-AGI-1GPT-5 by 765.758.7+7
- AA IntelligenceGPT-5 by 6.32316.7+6.3
- frontiermath_tier_4_v1GPT-5 by 6.212.56.3+6.2
- LiveCodeBenchGPT-5 by 6.183.577.4+6.1
- vectara_avg_summary_lengthGPT-5 by 35 rating points162.7127.7+35 rating
- vectara_factual_consistencyGPT-5 by 3.584.981.4+3.5
- bbqGPT-5 by 2.896.894+2.8
- AIME 2025GPT-5 by 1.994.692.7+1.9
- anthropic_red_teamtie99.198.2tie
- vectara_answer_ratetie99.999.2tie
- HarmBenchtie97.697tie
- XSTesttie97.197.4tie
- simple_safety_teststie99.8100tie
Questions people ask
Which is better, GPT-5 or o4-mini?
GPT-5 wins seven of the eight areas we test: coding, reasoning, facts, images and charts, math, long documents and following instructions. o4-mini wins none, but costs 51% less. They are level on agents.
Which is better for coding?
GPT-5. It wins 2 of the 3 coding tests both models report; o4-mini wins 1.
Which is cheaper?
GPT-5 costs $1.25 per million input tokens and $10.00 per million output tokens; o4-mini costs $1.10 and $4.40. That makes o4-mini about 51% 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 22 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 20 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.