Arena 멀티턴 Elo

이 벤치마크는?

여러 차례 대화가 이어지는 멀티턴 프롬프트에서의 사용자 선호도 기반 Arena Elo 점수입니다.

출처: Arena Intelligence
순위모델
#1

Anthropic

Claude Opus 4.6

1518
#2

Grok

Grok 4.20

1513
#3

Grok

Grok 4.20 (Reasoning)

1513
#4

Google

Gemini 3.1 Pro

1504
#5

OpenAI

GPT-5.4

1500
#6

OpenAI

GPT-5.4 Pro

1500
#7

Anthropic

Claude Opus 4.5

1487
#8

Google

Gemini 3 Flash

1483
#9

Z.ai

GLM-5.1

1479
#10

Anthropic

Claude Sonnet 4.6

1478
#11

OpenAI

GPT-5.4 Mini

1478
#12

Anthropic

Claude Opus 4.1

1473
#13

Anthropic

Claude Sonnet 4.5

1468
#14

Google

Gemma 4 31B

1463
#15

Z.ai

GLM-5

1461
#16

Xiaomi

MiMo-V2-Pro

1459
#17

Moonshot AI

Kimi K2.5

1455
#18

Google

Gemini 2.5 Pro

1452
#19

Qwen

Qwen3.5 397B A17B

1449
#20

Google

Gemini 3.1 Flash Lite

1447
#21

Baidu

ERNIE 5.0 Thinking

1441
#22

Anthropic

Claude Opus 4

1437
#23

Meituan

Longcat Flash Chat

1426
#24

DeepSeek

DeepSeek V3.2

1426
#25

Anthropic

Claude Haiku 4.5

1421
#26

OpenAI

GPT-5

1420
#27

Anthropic

Claude Sonnet 4

1419
#28

Grok

Grok 4.1 Fast

1416
#29

Grok

Grok 4.1 Fast (Reasoning)

1416
#30

MiniMax

MiniMax M2.5

1413
#31

OpenAI

GPT-5.4 Nano

1412
#32

MiniMax

MiniMax M2.7

1407
#33

Google

Gemini 2.5 Flash

1403
#34

Google

Gemini 2.5 Flash Lite

1374
#35

Arcee AI

Trinity Large Thinking

1372
#36

OpenAI

GPT-5 Mini

1372
#37

OpenAI

GPT OSS 120B

1328
#38

Meta AI

Llama 4 Maverick

1324
#39

Amazon

Nova 2 Lite

1324
#40

OpenAI

GPT-5 Nano

1323
#41

Meta AI

Llama 4 Scout

1320
#42

OpenAI

GPT-4.1

1298