Claude 4.5 Haiku (Reasoning) vs DeepSeek R1 Distill Llama 70B | grube.ai

Back to Benchmarks

Claude 4.5 Haiku (Reasoning)

Anthropic

vs

DeepSeek R1 Distill Llama 70B

DeepSeekvia DeepInfra

Claude 4.5 Haiku (Reasoning) wins 8 of 11

Index Scores

37.1

+21.1

Intelligence

16.0

32.6

+21.2

Coding

11.4

83.7

+30.0

Math

53.7

Benchmarks

76.0

MMLU Pro

+3.5

79.5

67.2

+27.0

GPQA

40.2

9.7

+3.6

HLE

6.1

61.5

+34.9

LiveCode

26.6

43.3

+12.1

SciCode

31.2

0.0

MATH-500

+93.5

93.5

0.0

AIME

+67.0

67.0

Speed

66

+24

TPS

42

Shared

Winning delta

crafted by bart stefanski