Compare · ModelsLive · 2 picked · head to head
GPT-4o (2024-11-20) vs GPT-4.5
Side by side · benchmarks, pricing, and signals you can act on.
Winner summary
GPT-4.5 wins on 10/11 benchmarks
GPT-4.5 wins 10 of 11 shared benchmarks. Leads in coding · reasoning · knowledge.
Category leads
coding·GPT-4.5reasoning·GPT-4.5knowledge·GPT-4.5math·GPT-4.5
Hype vs Reality
Attention vs performance
GPT-4o (2024-11-20)
#156 by perf·no signal
GPT-4.5
#166 by perf·no signal
Vendor risk
Who is behind the model
OpenAI
$840.0B·Tier 1
OpenAI
$840.0B·Tier 1
Head to head
11 benchmarks · 2 models
GPT-4o (2024-11-20)GPT-4.5
Aider polyglot
GPT-4.5 leads by +21.8
Aider Polyglot · measures how well AI models can edit code across multiple programming languages using the Aider coding assistant framework.
GPT-4o (2024-11-20)
23.1
GPT-4.5
44.9
ARC-AGI
GPT-4.5 leads by +5.8
ARC-AGI · the original Abstraction and Reasoning Corpus, testing whether AI can solve novel visual pattern recognition tasks without memorization.
GPT-4o (2024-11-20)
4.5
GPT-4.5
10.3
ARC-AGI-2
GPT-4.5 leads by +0.7
ARC-AGI-2 · the second iteration of the Abstraction and Reasoning Corpus, testing novel pattern recognition and abstract reasoning without prior training data.
GPT-4o (2024-11-20)
0.1
GPT-4.5
0.8
Cybench
GPT-4.5 leads by +5.0
Cybench · evaluates AI on real Capture-The-Flag cybersecurity challenges, testing vulnerability analysis, exploitation, and security reasoning.
GPT-4o (2024-11-20)
12.5
GPT-4.5
17.5
GPQA diamond
GPT-4.5 leads by +26.0
Graduate-Level Google-Proof QA (Diamond set) · expert-crafted questions in physics, biology, and chemistry that are difficult even for domain PhDs.
GPT-4o (2024-11-20)
32.3
GPT-4.5
58.3
Lech Mazur Writing
GPT-4o (2024-11-20) leads by +6.2
Lech Mazur Writing · evaluates creative writing ability, assessing prose quality, narrative coherence, and stylistic sophistication.
GPT-4o (2024-11-20)
81.8
GPT-4.5
75.6
MATH level 5
GPT-4.5 leads by +25.3
MATH Level 5 · the hardest tier of the MATH benchmark, featuring competition-level problems from AMC, AIME, and Olympiad-style mathematics.
GPT-4o (2024-11-20)
53.3
GPT-4.5
78.6
OTIS Mock AIME 2024-2025
GPT-4.5 leads by +31.4
OTIS Mock AIME 2024-2025 · simulated American Invitational Mathematics Examination problems testing advanced problem-solving skills.
GPT-4o (2024-11-20)
6.3
GPT-4.5
37.7
SimpleBench
GPT-4.5 leads by +20.0
SimpleBench · tests fundamental reasoning capabilities with straightforward problems designed to expose gaps in basic logical and spatial thinking.
GPT-4o (2024-11-20)
1.4
GPT-4.5
21.4
VPCT
GPT-4.5 leads by +7.5
VPCT (Visual Pattern Completion Test) · tests visual reasoning and pattern recognition by having models complete visual sequences and transformations.
GPT-4o (2024-11-20)
10.0
GPT-4.5
17.5
WeirdML
GPT-4.5 leads by +14.2
WeirdML · tests models on unusual and adversarial machine learning tasks that require creative problem-solving beyond standard patterns.
GPT-4o (2024-11-20)
25.1
GPT-4.5
39.4
Full benchmark table
| Benchmark | GPT-4o (2024-11-20) | GPT-4.5 |
|---|---|---|
Aider polyglot Aider Polyglot · measures how well AI models can edit code across multiple programming languages using the Aider coding assistant framework. | 23.1 | 44.9 |
ARC-AGI ARC-AGI · the original Abstraction and Reasoning Corpus, testing whether AI can solve novel visual pattern recognition tasks without memorization. | 4.5 | 10.3 |
ARC-AGI-2 ARC-AGI-2 · the second iteration of the Abstraction and Reasoning Corpus, testing novel pattern recognition and abstract reasoning without prior training data. | 0.1 | 0.8 |
Cybench Cybench · evaluates AI on real Capture-The-Flag cybersecurity challenges, testing vulnerability analysis, exploitation, and security reasoning. | 12.5 | 17.5 |
GPQA diamond Graduate-Level Google-Proof QA (Diamond set) · expert-crafted questions in physics, biology, and chemistry that are difficult even for domain PhDs. | 32.3 | 58.3 |
Lech Mazur Writing Lech Mazur Writing · evaluates creative writing ability, assessing prose quality, narrative coherence, and stylistic sophistication. | 81.8 | 75.6 |
MATH level 5 MATH Level 5 · the hardest tier of the MATH benchmark, featuring competition-level problems from AMC, AIME, and Olympiad-style mathematics. | 53.3 | 78.6 |
OTIS Mock AIME 2024-2025 OTIS Mock AIME 2024-2025 · simulated American Invitational Mathematics Examination problems testing advanced problem-solving skills. | 6.3 | 37.7 |
SimpleBench SimpleBench · tests fundamental reasoning capabilities with straightforward problems designed to expose gaps in basic logical and spatial thinking. | 1.4 | 21.4 |
VPCT VPCT (Visual Pattern Completion Test) · tests visual reasoning and pattern recognition by having models complete visual sequences and transformations. | 10.0 | 17.5 |
WeirdML WeirdML · tests models on unusual and adversarial machine learning tasks that require creative problem-solving beyond standard patterns. | 25.1 | 39.4 |
Pricing · per 1M tokens · projected $/mo at 10M tokens
| Model | Input | Output | Context | Projected $/mo |
|---|---|---|---|---|
| $2.50 | $10.00 | 128K tokens (~64 books) | $43.75 | |
| — | — | — | — |