Claude 4.5 Haiku (Reasoning) vs Llama 3.3 Instruct 70B | grube.ai

Back to Benchmarks

Claude 4.5 Haiku (Reasoning)

Anthropic

vs

Llama 3.3 Instruct 70B

Metavia Groq

Claude 4.5 Haiku (Reasoning) wins 8 of 11

Index Scores

37.1

+22.6

Intelligence

14.5

32.6

+21.9

Coding

10.7

83.7

+76.0

Math

7.7

Benchmarks

76.0

+4.7

MMLU Pro

71.3

67.2

+17.4

GPQA

49.8

9.7

+5.7

HLE

4.0

61.5

+32.7

LiveCode

28.8

43.3

+17.3

SciCode

26.0

0.0

MATH-500

+77.3

77.3

0.0

AIME

+30.0

30.0

Speed

66

TPS

+78

144

Shared

Winning delta

crafted by bart stefanski