Arena 수학 Elo

이 벤치마크는?

수학 관련 프롬프트에서의 사용자 선호도를 집계한 Arena Elo 점수입니다.

출처: Arena Intelligence
순위모델
#1

OpenAI

GPT-5.4

1522
#2

OpenAI

GPT-5.4 Pro

1522
#3

Anthropic

Claude Opus 4.6

1512
#4

Google

Gemini 3.1 Pro

1508
#5

Moonshot AI

Kimi K2.5

1478
#6

Google

Gemini 3 Flash

1476
#7

Google

Gemma 4 31B

1473
#8

Anthropic

Claude Opus 4.5

1471
#9

Anthropic

Claude Sonnet 4.6

1469
#10

Z.ai

GLM-5

1458
#11

Grok

Grok 4.20

1458
#12

Grok

Grok 4.20 (Reasoning)

1458
#13

Z.ai

GLM-5.1

1457
#14

Anthropic

Claude Sonnet 4.5

1454
#15

Qwen

Qwen3.5 397B A17B

1449
#16

Xiaomi

MiMo-V2-Pro

1448
#17

Baidu

ERNIE 5.0 Thinking

1448
#18

Meituan

Longcat Flash Chat

1447
#19

Google

Gemini 2.5 Pro

1444
#20

Anthropic

Claude Opus 4.1

1443
#21

Google

Gemini 3.1 Flash Lite

1438
#22

OpenAI

GPT-5

1433
#23

DeepSeek

DeepSeek V3.2

1429
#24

Grok

Grok 4.1 Fast

1419
#25

Grok

Grok 4.1 Fast (Reasoning)

1419
#26

OpenAI

GPT-5.4 Nano

1419
#27

Anthropic

Claude Opus 4

1418
#28

MiniMax

MiniMax M2.5

1418
#29

OpenAI

GPT-5.4 Mini

1414
#30

Google

Gemini 2.5 Flash

1409
#31

MiniMax

MiniMax M2.7

1406
#32

OpenAI

GPT-5 Mini

1405
#33

Anthropic

Claude Sonnet 4

1402
#34

Anthropic

Claude Haiku 4.5

1392
#35

OpenAI

GPT OSS 120B

1384
#36

Google

Gemini 2.5 Flash Lite

1365
#37

Arcee AI

Trinity Large Thinking

1362
#38

OpenAI

GPT-5 Nano

1342
#39

Amazon

Nova 2 Lite

1335
#40

Meta AI

Llama 4 Maverick

1319
#41

Meta AI

Llama 4 Scout

1310
#42

OpenAI

GPT-4.1

1302