Noometry

Model comparison

GPT-4 vs GPT-4.5

GPT-4.5 is the stronger model overall, scoring 37.2 to 29.1 on the Noometry Index.

Last verified . 23 shared benchmarks.

GPT-4 OpenAI

29.1

Rank #316 Confirmed

GPT-4.5 OpenAI

37.2

Rank #208 Confirmed

Summary

  • They share 23 benchmarks with published results for both. GPT-4 scores higher in 1 category and GPT-4.5 in 7 categories; 8 gaps are clear of the uncertainty.
  • The widest gap is in writing & preference, where GPT-4.5 leads 56.9 to 34.9.
  • The biggest single-benchmark swing is MATH Level 5: 23% for GPT-4 and 78.6% for GPT-4.5.

Side by side

GPT-4GPT-4.5
ProviderOpenAIOpenAI
Noometry Index29.137.2
Released2023-03-142025-02-27
WeightsProprietaryProprietary
Context window8K—
Max output8K—
Input $ / M tokens$30—
Output $ / M tokens$60—
Results tracked3842

Sponsored placements are available on pages like this one. Advertise on Noometry

Category by category

Coding GPT-4.5 leads

GPT-4: 31.6 (#283), GPT-4.5: 42.2 (#109)

BenchmarkGPT-4GPT-4.5
WeirdML12.4%39.4%
LMArena Coding12541396
Aider Polyglot—44.9%
BigCodeBench Instruct46%—
LiveBench Coding—75.2%
BigCodeBench Complete57.2%—
HumanEval+79.3%—

Agentic & Tool Use Not comparable

GPT-4: —, GPT-4.5: 27.9 (#97)

BenchmarkGPT-4GPT-4.5
Cybench—17.5%
METR Time Horizons36.1%—

Reasoning GPT-4 leads

GPT-4: 17.8 (#289), GPT-4.5: 13.9 (#330)

BenchmarkGPT-4GPT-4.5
LMArena Hard Prompts12411403
Epoch Capabilities Index125.89136.74
ForecastBench57.861.7
ARC-AGI-2—0.8%
SimpleBench—34.5%
ARC-AGI-1—10.3%
Chess Puzzles4%—
EnigmaEval—3.2%
LiveBench Reasoning—71.1%
Mystery Game Puzzles12%—
DTBench62.7%—
LiveBench Data Analysis—64.3%
LMCA17.1%—
BIG-Bench Hard75.1%—
HellaSwag95.3%—
LiveBench—69%
WinoGrande87.5%—

Math GPT-4.5 leads

GPT-4: 10.8 (#309), GPT-4.5: 32.6 (#211)

BenchmarkGPT-4GPT-4.5
OTIS Mock AIME 2024-20251.1%37.8%
LMArena Math12691412
MATH Level 523%78.6%
LiveBench Math—69.3%
GSM8K92%—

Knowledge GPT-4.5 leads

GPT-4: 18.4 (#282), GPT-4.5: 32.5 (#211)

BenchmarkGPT-4GPT-4.5
GPQA Diamond35.7%68.7%
LMArena Expert12111394
Humanity's Last Exam—5.4%
Confabulations—13.6%
MMLU86.4%—
TriviaQA84.8%—

Multimodal Not comparable

GPT-4: —, GPT-4.5: 37.6 (#71)

BenchmarkGPT-4GPT-4.5
LMArena Vision—1195
VPCT—45%

Multilingual GPT-4.5 leads

GPT-4: 40.6 (#215), GPT-4.5: 52.5 (#83)

BenchmarkGPT-4GPT-4.5
LMArena Non-English12461413
LMArena Chinese12421421
LMArena French12831418
LMArena German12511457
LMArena Japanese12091416
LMArena Korean11841392
LMArena Russian12511419
LMArena Spanish1261—

Instruction Following GPT-4.5 leads

GPT-4: 65.3 (#222), GPT-4.5: 72.6 (#134)

BenchmarkGPT-4GPT-4.5
LMArena Instruction Following12411404
LiveBench Instruction Following—72.3%

Long Context GPT-4.5 leads

GPT-4: 37.7 (#212), GPT-4.5: 40.4 (#155)

BenchmarkGPT-4GPT-4.5
LMArena Longer Query12441406
Fiction.LiveBench—63.9%

Writing & Preference GPT-4.5 leads

GPT-4: 34.9 (#268), GPT-4.5: 56.9 (#134)

BenchmarkGPT-4GPT-4.5
LMArena Text12631417
LMArena Creative Writing12441394
EQ-Bench Creative Writing7521258
LMArena Multi-Turn12571444
Short-Story Creative Writing—75.6%
LiveBench Language—61.5%

Frequently asked questions

Is GPT-4 better than GPT-4.5?

GPT-4.5 is the stronger model overall, scoring 37.2 to 29.1 on the Noometry Index.

Is GPT-4 or GPT-4.5 better for coding?

GPT-4.5 scores higher on coding benchmarks: 42.2 versus 31.6 in the Noometry coding category.

How many benchmarks do GPT-4 and GPT-4.5 share?

23 benchmarks have published results for both models. GPT-4 has 38 scored results on Noometry and GPT-4.5 has 42.

Related comparisons

Go deeper