← Arena leaderboards

Text · Factuality · Japanese

Arena · 50 models

RankModelScore
1Claude Opus 5 (High)1,517.9524
2Gemini 3.8 Flash (High)1,512.4349
3GPT-5.5 (High)1,505.3083
4Kimi K3 (Max)1,495.0764
5GPT-5.51,490.8167
6GPT-5.4 (High)1,483.632
7Claude Fable 5 (High)1,483.289
8Gemini 3.1 Pro Preview1,482.0733
9Claude Opus 4.6 (High)1,478.1011
10Claude Opus 4.7 (High)1,476.645
11Claude Opus 4.61,473.3449
12Claude Opus 4.71,468.4012
13GPT-5.41,466.3032
14DeepSeek-V4-Pro1,461.8518
15Gemini 3.5 Flash (Medium)1,458.0656
16Claude Opus 4.81,457.106
17GPT-5.6 Terra (xHigh)1,456.4037
18Claude Sonnet 4.61,454.0382
19GLM-5.3-Flash1,453.9765
20GLM-5.2 (Max)1,449.0521
21Muse Spark 1.11,444.6133
22Claude Sonnet 5 (High)1,442.754
23Claude Opus 4.8 (High)1,441.7394
24Kimi K2.61,441.4165
25GLM-5.11,440.7443
26Qwen3.5-397B-A17B1,436.6625
27DeepSeek-V4-Pro Preview (High)1,431.1627
28Grok 4.20 Beta (Reasoning, 0309)1,426.7286
29Grok 4.20 Multi-Agent Beta (0309)1,426.4411
30Gemini 3 Flash (Minimal)1,424.6695
31Gemini 3.1 Flash-Lite Preview1,420.9299
32GPT-5.21,419.2893
33ERNIE 5.11,419.0417
34Kimi K2.5 (Thinking)1,417.8334
35DeepSeek-V4-Flash Preview (High)1,415.0336
36MiMo-V2.5-Pro1,414.0329
37Grok 4.31,413.1158
38Qwen3.7-Plus1,411.6132
39DeepSeek-V4-Flash1,410.0818
40GPT-6 Luna (Max)1,405.6667
41Mistral Large 31,399.0674
42GPT-5.4 mini (High)1,396.4346
43GPT-5.4 nano (High)1,392.5275
44Qwen3.6-Plus1,390.3983
45Dola Seed 2.0 Pro1,389.5047
46MiniMax-M31,385.4594
47Claude Haiku 4.5 (20251001)1,384.3787
48Qwen3.5-Flash1,383.4433
49Step-3.5-Flash1,370.0706
50MiMo-V2.51,349.3847