GPT-4o vs GPT-4o mini
Wins 4 of 6 areas
Coding · Facts · Images and charts · Following instructions
Wins 0 of 6 areas
—
GPT-4o is the stronger all-rounder.GPT-4o mini is cheaper.
Scores updated · 53 tests both models report · How we compare
Where each one wins
Tests won in each of the six areas where both have results. Each piece is one test, so a longer bar means more evidence; grey means the two scored within a point of each other.
- Images and chartsUnderstanding pictures, charts and video30GPT-4o3 of 3 tests
- CodingWriting and fixing software20GPT-4o2 of 2 tests
- FactsGetting facts right instead of making them up10GPT-4o1 of 1 test
- Following instructionsDoing exactly what it is asked10GPT-4o1 of 1 test
- AgentsCarrying out multi-step tasks on its own00Even0 each · 1 tie
- ReasoningHard problems that need careful thinking11Even1 each · 1 tie
Agents, facts 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.
GPT-4o mini costs 96% 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 GPT-4o pulls ahead
- Fixes real GitHub issues in Python projectsSWE-bench Verified+24.5points ahead
- Short factual questions, answered correctlySimpleQA Verified+17.7points ahead
- Harder college exam questions with imagesMMMU-Pro+14.8points ahead
Where GPT-4o mini pulls ahead
- Very hard expert questions across many subjectsHumanity's Last Exam+2.4points ahead
Every test, side by side
All 53 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingGPT-4o
- SWE-bench VerifiedGPT-4o by 24.533.28.7+24.5
- SciCodeGPT-4o by 10.433.322.9+10.4
ReasoningEven
- GPQA DiamondGPT-4o by 1052.642.6+10
- Humanity's Last ExamGPT-4o mini by 2.41.84.2+2.4
- ARC-AGI-2tie00tie
Images and chartsGPT-4o
Other results42 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.
- SimpleQAGPT-4o by 28.338.29.9+28.3
- Aider-PolyglotGPT-4o by 27.130.73.6+27.1
- InfoVQA (Acc)GPT-4o by 22.880.757.9+22.8
- AI2DGPT-4o by 1994.275.2+19
- MMMU (val) (Pass@1)GPT-4o by 1769.152.1+17
- MLVU-MCQ (Acc)GPT-4o by 16.564.648.1+16.5
- BLINK (Acc)GPT-4o by 14.46853.6+14.4
- MMLongBench-DOC (Acc)GPT-4o by 13.842.829+13.8
- MMLU-ProGPT-4o by 1374.761.7+13
- Artificial Analysis Coding IndexGPT-4o by 12.824.211.4+12.8
- ScreenSpot-V2GPT-4o by 11.218.16.9+11.2
- naturalquestions_closedbookGPT-4o by 1149.638.5+11
- MMStar (Acc)GPT-4o by 9.964.754.8+9.9
- TOMATOGPT-4o by 8.937.728.8+8.9
- LongVideoBench (val)GPT-4o by 8.566.758.2+8.5
- RealWorldQA (Acc)GPT-4o by 8.375.467.1+8.3
- VideoMME (w sub.)GPT-4o by 8.377.268.9+8.3
- Video-MME (w/o sub.)GPT-4o by 7.171.964.8+7.1
- MMLUGPT-4o by 773.866.8+7
- livecodebench_mediumGPT-4o by 6.932.125.2+6.9
- AI2D (Acc)GPT-4o by 6.884.677.8+6.8
- WindowsAgentArena (Pass@1)GPT-4o by 6.79.42.7+6.7
- GSM8KGPT-4o by 6.690.984.3+6.6
- AIR-Bench 2024GPT-4o by 6.162.456.3+6.1
- MMBench-EN-v1.1 (Acc)GPT-4o by 683.177.1+6
- MMVU-Val (Pass@1)GPT-4o by 5.867.461.6+5.8
- MATHGPT-4o by 5.185.380.2+5.1
- OpenBookQAGPT-4o by 4.896.892+4.8
- DROPGPT-4o by 3.783.479.7+3.7
- MGSMGPT-4o by 3.590.587+3.5
- HumanEvalGPT-4o by 390.287.2+3
- OCRBench (Acc)GPT-4o by 30 rating points815785+30 rating
- LiveCodeBenchGPT-4o by 2.838.335.5+2.8
- NarrativeQAGPT-4o by 2.779.576.8+2.7
- MMVet (Pass@1)GPT-4o by 2.269.166.9+2.2
- HarmBenchGPT-4o mini by 282.984.9+2
- XSTestGPT-4o by 1.397.396+1.3
- anthropic_red_teamtie99.198.3tie
- livecodebench_easytie82.581.7tie
- simple_safety_teststie98.597.8tie
- AA Intelligencetie7.36.7tie
- livecodebench_hardtie4.54.3tie
Questions people ask
Which is better, GPT-4o or GPT-4o mini?
GPT-4o wins four of the six areas where both have results: coding, facts, images and charts and following instructions. GPT-4o mini wins none, but costs 96% less. They are level on agents and reasoning.
Which is better for coding?
GPT-4o. It wins 2 of the 2 coding tests both models report; GPT-4o mini wins none.
Which is cheaper?
GPT-4o costs $5.00 per million input tokens and $15.00 per million output tokens; GPT-4o mini costs $0.15 and $0.60. That makes GPT-4o mini about 96% cheaper for the same work.
How do you compare the two?
We use the 53 benchmark tests both models have published scores on. The verdict counts the 11 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 42 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.