SciCode

이 벤치마크는?

수학, 물리, 화학, 생물학 등 과학 분야의 연구 수준 코딩 문제로 구성된 벤치마크. 실제 과학 연구에서 사용되는 수준의 코드 작성 능력을 평가합니다. 점수는 정답률(%)입니다.

출처: Artificial Analysis
순위모델
#1

Google

Gemini 3.1 Pro

58.9%
#2

OpenAI

GPT-5.4

56.6%
#3

Anthropic

Claude Opus 4.6

51.9%
#4

Meta AI

Muse Spark

51.5%
#5

Google

Gemini 3 Flash

50.6%
#6

Anthropic

Claude Opus 4.5

49.5%
#7

Moonshot AI

Kimi K2.5

49.0%
#8

MiniMax

MiniMax M2.7

47.0%
#9

Z.ai

GLM-5

46.2%
#10

Grok

Grok 4.20 (Reasoning)

45.6%
#11

Anthropic

Claude Sonnet 4.5

44.7%
#12

OpenAI

GPT-5.4 Mini

44.2%
#13

Grok

Grok 4.1 Fast (Reasoning)

44.2%
#14

Anthropic

Claude Sonnet 4.6

44.1%
#15

Google

Gemma 4 31B

43.4%
#16

Anthropic

Claude Haiku 4.5

43.3%
#17

Google

Gemini 2.5 Pro

42.8%
#18

MiniMax

MiniMax M2.5

42.6%
#19

Xiaomi

MiMo-V2-Pro

42.5%
#20

Google

Gemini 3.1 Flash Lite

41.9%
#21

Qwen

Qwen3.5 397B A17B

41.1%
#22

Anthropic

Claude Opus 4.1

40.9%
#23

Qwen

Qwen3.6 Plus

40.7%
#24

Anthropic

Claude Sonnet 4

40.0%
#25

Anthropic

Claude Opus 4

39.8%
#26

Google

Gemini 2.5 Flash

39.4%
#27

DeepSeek

DeepSeek V3.2

38.9%
#28

OpenAI

GPT-5.4 Nano

38.4%
#29

OpenAI

GPT-4.1

38.1%
#30

OpenAI

GPT-5

37.8%
#31

Baidu

ERNIE 5.0 Thinking

37.5%
#32

OpenAI

GPT-5 Mini

36.9%
#33

OpenAI

GPT OSS 120B

36.0%
#34

NVIDIA

Nemotron 3 Super

36.0%
#35

OpenAI

GPT-5 Nano

33.8%
#36

Meta AI

Llama 4 Maverick

33.1%
#37

Grok

Grok 4.20

32.8%
#38

Baidu

ERNIE 4.5 300B A47B

31.5%
#39

Grok

Grok 4.1 Fast

29.6%
#40

Google

Gemini 2.5 Flash Lite

28.5%
#41

Meituan

Longcat Flash Chat

28.4%
#42

Amazon

Nova 2 Lite

24.0%
#43

Meta AI

Llama 4 Scout

17.0%
#44

Mistral AI

Mistral Small 4

13.4%