Arena Elo score aggregated from user preference votes on Korean-language prompts.
Source: Arena Intelligence| Rank | Model | |
|---|---|---|
| #1 | Gemini 3.1 Pro | 1487 |
| #2 | Anthropic Claude Sonnet 4.6 | 1453 |
| #3 | Gemini 3 Flash | 1451 |
| #4 | Z.ai GLM-5 | 1431 |
| #5 | Anthropic Claude Opus 4.6 | 1422 |
| #6 | Gemini 3.1 Flash Lite | 1420 |
| #7 | Anthropic Claude Opus 4.1 | 1417 |
| #8 | Gemini 2.5 Pro | 1417 |
| #9 | Xiaomi MiMo-V2-Pro | 1416 |
| #10 | Anthropic Claude Opus 4.5 | 1407 |
| #11 | ChatGPT GPT-5.4 | 1406 |
| #12 | ChatGPT GPT-5.4 Pro | 1406 |
| #13 | Baidu ERNIE 5.0 Thinking | 1405 |
| #14 | Moonshot AI Kimi K2.5 | 1405 |
| #15 | ChatGPT GPT-5 | 1400 |
| #16 | Grok Grok 4.1 Fast | 1388 |
| #17 | Grok Grok 4.1 Fast (Reasoning) | 1388 |
| #18 | Anthropic Claude Opus 4 | 1381 |
| #19 | Anthropic Claude Sonnet 4.5 | 1380 |
| #20 | Gemini 2.5 Flash | 1376 |
| #21 | DeepSeek DeepSeek V3.2 | 1366 |
| #22 | Qwen Qwen3.5 397B A17B | 1361 |
| #23 | Anthropic Claude Haiku 4.5 | 1347 |
| #24 | Gemini 2.5 Flash Lite | 1344 |
| #25 | Anthropic Claude Sonnet 4 | 1336 |
| #26 | Arcee AI Trinity Large Thinking | 1330 |
| #27 | ChatGPT GPT-5 Mini | 1324 |
| #28 | MiniMax MiniMax M2.5 | 1300 |
| #29 | ChatGPT GPT-5 Nano | 1273 |
| #30 | Meta AI Llama 4 Scout | 1273 |
| #31 | Meta AI Llama 4 Maverick | 1264 |
| #32 | ChatGPT GPT OSS 120B | 1263 |
| #33 | ChatGPT GPT-4.1 | 1255 |
| #34 | Amazon Nova 2 Lite | 1229 |