GPT-4.5 Preview vs GPT-4o
Wins 4 of 4 areas
Coding · Reasoning · Images and charts · Following instructions
Wins 0 of 4 areas
—
GPT-4.5 Preview is the stronger all-rounder.
Scores updated · 22 tests both models report · How we compare
Where each one wins
Tests won in each of the four 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 video21GPT-4.5 Preview2 of 3 tests
- CodingWriting and fixing software10GPT-4.5 Preview1 of 1 test
- ReasoningHard problems that need careful thinking10GPT-4.5 Preview1 of 1 test
- Following instructionsDoing exactly what it is asked10GPT-4.5 Preview1 of 1 test
Coding, reasoning 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.
The biggest differences
The tests each model wins by the widest margin, up to three each. Scores are out of 100.
Where GPT-4.5 Preview pulls ahead
- Graduate-level biology, physics and chemistry questionsGPQA Diamond+16.9points ahead
- Math problems shown in pictures and chartsMathVista+10.9points ahead
- Fixes real GitHub issues in Python projectsSWE-bench Verified+4.8points ahead
Where GPT-4o pulls ahead
- Reasoning about charts from research papersCharXiv (RQ)+3.4points ahead
Every test, side by side
All 22 tests both models report. The winning score is in its model's colour; marks a score checked independently.
Images and chartsGPT-4.5 Preview
- MathVistaGPT-4.5 Preview by 10.972.361.4+10.9
- CharXiv (RQ)GPT-4o by 3.455.458.8+3.4
- MMMUGPT-4.5 Preview by 375.272.2+3
Following instructionsGPT-4.5 Preview
- Multi-ChallengeGPT-4.5 Preview by 3.543.840.3+3.5
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.
- SimpleQAGPT-4.5 Preview by 24.362.538.2+24.3
- Aider-PolyglotGPT-4.5 Preview by 14.244.930.7+14.2
- HarmBenchGPT-4.5 Preview by 12.995.882.9+12.9
- AIR-Bench 2024GPT-4.5 Preview by 11.874.262.4+11.8
- TAU-bench (retail)GPT-4.5 Preview by 8.168.460.3+8.1
- IFEvalGPT-4.5 Preview by 7.288.281+7.2
- TAU-bench (airline)GPT-4.5 Preview by 7.25042.8+7.2
- GSM8KGPT-4.5 Preview by 6.19790.9+6.1
- ARC-AGI-1GPT-4.5 Preview by 5.810.34.5+5.8
- MMMLUGPT-4.5 Preview by 3.785.181.4+3.7
- bbqGPT-4o by 3.19295.1+3.1
- AA IntelligenceGPT-4.5 Preview by 2.39.67.3+2.3
- HumanEvalGPT-4o by 2.28890.2+2.2
- XSTestGPT-4o by 2.295.197.3+2.2
- simple_safety_testsGPT-4.5 Preview by 1.510098.5+1.5
- anthropic_red_teamtie99.499.1tie
Questions people ask
Which is better, GPT-4.5 Preview or GPT-4o?
GPT-4.5 Preview wins all four areas where both have results: coding, reasoning, images and charts and following instructions. GPT-4o wins none.
Which is better for coding?
GPT-4.5 Preview. It wins the one coding test both models report.
How do you compare the two?
We use the 22 benchmark tests both models have published scores on. The verdict counts the 6 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.