← Arena leaderboards

Vision · Style Control · Captioning

Arena · 34 models

RankModelScore
1Gemini 3.1 Pro Preview1,281.6156
2Gemini 3 Pro1,276.5851
3GPT-5.2 (High)1,262.1194
4Gemini 2.5 Pro1,250.5864
5GPT-5.1 (High)1,248.237
6Gemma 4 26B A4B1,238.9268
7Gemini 3 Flash1,233.7928
8Gemini 3.1 Flash-Lite Preview1,225.4807
9ChatGPT-4o Latest (20250326)1,212.7463
10Gemini 2.5 Flash1,211.3501
11Qwen3-VL-235B-A22B-Instruct1,211.0336
12GPT-4.1 (2025-04-14)1,209.9203
13Gemini 3 Flash (Minimal)1,209.0135
14Kimi K2.5 (Thinking)1,197.9607
15GPT-5 (High)1,196.0498
16GPT-5 Chat1,194.2233
17GPT-4.1 mini (2025-04-14)1,193.2599
18o4-mini (2025-04-16)1,184.8136
19o3 (2025-04-16)1,182.5713
20Gemini 2.5 Flash-Lite Preview (Thinking, 06-17)1,182.4442
21GLM-5V-Turbo1,179.7555
22GPT-5 mini (High)1,175.1711
23Qwen3.5-397B-A17B1,174.4849
24GPT-5.21,171.8627
25Grok 4 (0709)1,170.472
26Gemma 4 31B1,170.0801
27GPT-5.11,162.6467
28Gemini 2.0 Flash (001)1,149.8662
29Mistral Small 3.1 24B Instruct (2503)1,133.0815
30Mistral Medium 3.1 (2508)1,129.551
31Gemma 3 27B IT1,116.7484
32Qwen3.5-27B1,095.4517
33Mistral Medium 3 (2505)1,090.0444
34Mistral Small 3.2 (2506)1,050.491