Noometry

Model comparison

Claude 3 Haiku vs GPT-4

GPT-4 is the stronger model overall, scoring 28.9 to 25.7 on the Noometry Index.

Last verified . 31 shared benchmarks.

Claude 3 Haiku Anthropic

25.7

Rank #333 Confirmed

GPT-4 OpenAI

28.9

Rank #308 Confirmed

Summary

  • They share 31 benchmarks with published results for both. Claude 3 Haiku scores higher in 0 categories and GPT-4 in 8 categories; 7 gaps are clear of the uncertainty.
  • The widest gap is in writing & preference, where GPT-4 leads 35.0 to 29.8.
  • The biggest single-benchmark swing is DTBench: 50.1% for Claude 3 Haiku and 62.7% for GPT-4.

Side by side

Claude 3 HaikuGPT-4
ProviderAnthropicOpenAI
Noometry Index25.728.9
Released2024-03-072023-03-14
WeightsProprietaryProprietary
Context window—8K
Max output—8K
Input $ / M tokens—$30
Output $ / M tokens—$60
Results tracked3738

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

Category by category

Coding GPT-4 leads

Claude 3 Haiku: 26.1 (#320), GPT-4: 31.2 (#277)

BenchmarkClaude 3 HaikuGPT-4
WeirdML9.8%12.4%
BigCodeBench Instruct39.4%46%
LMArena Coding11991254
BigCodeBench Complete50.1%57.2%
HumanEval+68.9%79.3%
CadEval12%—
MBPP+68.8%—

Agentic & Tool Use Not comparable

Claude 3 Haiku: —, GPT-4: —

BenchmarkClaude 3 HaikuGPT-4
METR Time Horizons—36.1%

Reasoning GPT-4 leads

Claude 3 Haiku: 15.7 (#303), GPT-4: 17.3 (#285)

BenchmarkClaude 3 HaikuGPT-4
LMArena Hard Prompts11741241
DTBench50.1%62.7%
LMCA8.8%17.1%
Epoch Capabilities Index118.3125.88
ForecastBench53.257.8
WinoGrande74.2%87.5%
Kagi LLM Benchmark34.2%—
Chess Puzzles—4%
Mystery Game Puzzles—12%
BIG-Bench Hard—75.1%
HellaSwag—95.3%

Math Too close to call

Claude 3 Haiku: 9.2 (#313), GPT-4: 10.1 (#304)

BenchmarkClaude 3 HaikuGPT-4
OTIS Mock AIME 2024-20251.8%1.1%
LMArena Math11881269
MATH Level 514.9%23%
GSM8K—92%

Knowledge GPT-4 leads

Claude 3 Haiku: 17.3 (#280), GPT-4: 18.4 (#277)

BenchmarkClaude 3 HaikuGPT-4
GPQA Diamond36.3%35.7%
LMArena Expert11481211
MMLU73.8%86.4%
Confabulations34.2%—
TriviaQA—84.8%

Multimodal Not comparable

Claude 3 Haiku: 23.1 (#123), GPT-4: —

BenchmarkClaude 3 HaikuGPT-4
LMArena Vision950—
ScienceQA72%—

Multilingual GPT-4 leads

Claude 3 Haiku: 36.1 (#238), GPT-4: 40.8 (#210)

BenchmarkClaude 3 HaikuGPT-4
LMArena Non-English11781246
LMArena Chinese11551242
LMArena French11941282
LMArena German11741251
LMArena Japanese11021209
LMArena Korean11081184
LMArena Russian12041251
LMArena Spanish11651261

Instruction Following GPT-4 leads

Claude 3 Haiku: 61.5 (#242), GPT-4: 65.5 (#217)

BenchmarkClaude 3 HaikuGPT-4
LMArena Instruction Following11741241

Long Context GPT-4 leads

Claude 3 Haiku: 36.1 (#232), GPT-4: 37.7 (#207)

BenchmarkClaude 3 HaikuGPT-4
LMArena Longer Query11911244

Writing & Preference GPT-4 leads

Claude 3 Haiku: 29.8 (#286), GPT-4: 35.0 (#263)

BenchmarkClaude 3 HaikuGPT-4
LMArena Text11951263
LMArena Creative Writing11571244
EQ-Bench Creative Writing717752
LMArena Multi-Turn11901257

Frequently asked questions

Is Claude 3 Haiku better than GPT-4?

GPT-4 is the stronger model overall, scoring 28.9 to 25.7 on the Noometry Index.

Is Claude 3 Haiku or GPT-4 better for coding?

GPT-4 scores higher on coding benchmarks: 31.2 versus 26.1 in the Noometry coding category.

How many benchmarks do Claude 3 Haiku and GPT-4 share?

31 benchmarks have published results for both models. Claude 3 Haiku has 37 scored results on Noometry and GPT-4 has 38.

Related comparisons

Go deeper