Design Arena · 游戏开发
定期获取并展示公开的基准测试资料(LMArena 用户投票排名、Epoch AI 测试分数、OpenRouter 使用量排名)。每个榜单均标注数据发布日期和获取日期。这些分数并非本站自行评定。
如何解读此排行榜
衡量什么: 这是 Design Arena 中,人们比较根据相同请求生成的简单游戏并投票得出的排名。
如何解读分数: 采用 Elo 方式计算相对得分,得分越高,表示人们更经常偏好该模型的结果。同时显示胜率。
注意事项: 主要评测简短的单个游戏,因此可能与大型项目开发能力不同。
本次排名的解读要点
- 最高分为 Claude Opus 5.5 的 1479。
- 前 10 个模型中,Anthropic 的模型有 4 个,数量最多。
游戏开发 Design Arena
发布日期:2026-10-05 · 获取日期:2026-10-05
图表仅显示同一模型各推理设置(high、max 等)中排名最高的一项。表格列出所有设置各自的排名。
查看表格(前 60 名)
| 排名 | 模型 | 开发方 | 分数 | 胜率 |
|---|---|---|---|---|
| 1 | #1Claude Opus 5.5 | Anthropic | 1479 | 73.6% |
| 2 | #2Claude Fable 5.1 | Anthropic | 1394 | 65.6% |
| 3 | #3Kimi K3 | Moonshot AI | 1379 | 61.8% |
| 4 | #4Muse Spark 1.3 Max | Meta | 1358 | 56.2% |
| 5 | #5Claude Opus 5 | Anthropic | 1350 | 59.1% |
| 6 | #6Claude Fable 5 | Anthropic | 1349 | 60.8% |
| 7 | #7Muse Spark 1.3 (xhigh) | Meta | 1343 | 54.6% |
| 8 | #8Gemini 3.7 Flash | 1335 | 54.2% | |
| 9 | #9ZGLM-5.3 | Z.ai | 1334 | 54.6% |
| 10 | #10Qwen3.8 Max | Alibaba | 1329 | 53.3% |
| 11 | #11MiMo-V2.6-Pro | Xiaomi | 1323 | 53% |
| 12 | #12Gemini 3.8 Flash | 1320 | 51.3% | |
| 13 | #13Muse Spark 1.2 | Meta | 1314 | 50.4% |
| 14 | #14OGPT-5.5 | OpenAI | 1308 | 58.1% |
| 15 | #15xGrok 4.6 | xAI | 1307 | 54.2% |
| 16 | #16xGrok 4.7 | xAI | 1304 | 51.5% |
| 17 | #17Claude Sonnet 5 | Anthropic | 1296 | 53% |
| 18 | #18ZGLM-5.3-Flash | Z.ai | 1294 | 47.2% |
| 19 | #19Claude Opus 4.6 (Thinking) | Anthropic | 1290 | 61% |
| 20 | #20Claude Opus 4.6 | Anthropic | 1285 | 60.5% |
| 21 | #21Muse Spark 1.1 | Meta | 1285 | 48.1% |
| 22 | #22ZGLM 5.2 | Z.ai | 1281 | 49.5% |
| 23 | #23Qwen3.7 Max | Alibaba | 1280 | 53.6% |
| 24 | #24Claude Opus 4.8 | Anthropic | 1275 | 53.2% |
| 25 | #25ZGLM 5.1 | Z.ai | 1275 | 55.6% |
| 26 | #26Gemini 3.6 Flash | 1275 | 51.1% | |
| 27 | #27xGrok 4.5 | xAI | 1275 | 49.3% |
| 28 | #28Gemini 3.5 Flash | 1274 | 53% | |
| 29 | #29Qwen3.7 Plus | Alibaba | 1273 | 49.7% |
| 30 | #30Claude Sonnet 4.6 | Anthropic | 1272 | 58.8% |
| 31 | #31MiMo-V2.5-Pro | Xiaomi | 1272 | 53.4% |
| 32 | #32Kimi K2.6 | Moonshot AI | 1263 | 54.9% |
| 33 | #33ZGLM 5 Turbo | Z.ai | 1260 | 53.2% |
| 34 | #34MiMo-V2.5 | Xiaomi | 1257 | 52.6% |
| 35 | #35OGPT-5.4 (Medium) | OpenAI | 1251 | 57.6% |
| 36 | #36ZGLM 5 | Z.ai | 1247 | 57.4% |
| 37 | #37Claude Opus 4.5 | Anthropic | 1242 | 59.3% |
| 38 | #38DeepSeek-V4-Pro | DeepSeek | 1242 | 52.3% |
| 39 | #39NNex N2 Pro | Nex Agi | 1238 | 49% |
| 40 | #40Kimi K2.7 Code | Moonshot AI | 1233 | 47.5% |
| 41 | #41ZGLM 5V Turbo | Z.ai | 1231 | 50.7% |
| 42 | #42MiniMax M3 | MiniMax | 1225 | 46% |
| 43 | #43Qwen3.6 Plus | Alibaba | 1224 | 49.1% |
| 44 | #44MiniMax M2.7 | MiniMax | 1223 | 50% |
| 45 | #45DeepSeek-V4-Flash-0731 | DeepSeek | 1221 | 44.6% |
| 46 | #46Kimi K2.5 (Thinking) | Moonshot AI | 1219 | 53.4% |
| 47 | #47Gemini 3.1 Pro Preview | 1215 | 52.1% | |
| 48 | #48DeepSeek-V4-Flash | DeepSeek | 1210 | 50.2% |
| 49 | #49Gemini 3 Pro Preview | 1209 | 60.9% | |
| 50 | #50OGPT-5.2 (Low) | OpenAI | 1208 | 56% |
| 51 | #51OGPT-5.4 (None) | OpenAI | 1207 | 53.6% |
| 52 | #52xGrok 4.20 Beta (Reasoning) | xAI | 1205 | 50.1% |
| 53 | #53OGPT-5.4 (Low) | OpenAI | 1204 | 52.8% |
| 54 | #54Claude 3.7 Sonnet | Anthropic | 1203 | 54.3% |
| 55 | #55ZGLM 4.7 | Z.ai | 1201 | 55.2% |
| 56 | #56OGPT-5.2 (High) | OpenAI | 1199 | 56.7% |
| 57 | #57OGPT-5 (High) | OpenAI | 1198 | 59.3% |
| 58 | #58MiMo-V2-Flash | Xiaomi | 1197 | 48.6% |
| 59 | #59OGPT-5.2 (Medium) | OpenAI | 1196 | 52.8% |
| 60 | #60OGPT-5.1 (High) | OpenAI | 1191 | 55.9% |
Source: Design Arena (www.designarena.ai) via OpenRouter (openrouter.ai/rankings). Licensed under CC BY 4.0.
数据:LMArena leaderboard dataset(CC BY 4.0)。修改:各榜单仅显示前 50 名,分数四舍五入。 https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset公司标志为各公司的商标,仅用于帮助区分(图标:Simple Icons)。数据:Epoch AI — Capabilities & benchmarking(CC BY 4.0)。 https://epoch.ai/benchmarksSource: Artificial Analysis (artificialanalysis.ai) via OpenRouter (openrouter.ai/rankings). Source: OpenRouter evals (openrouter.ai) via OpenRouter (openrouter.ai/rankings). Source: Design Arena (www.designarena.ai) via OpenRouter (openrouter.ai/rankings). Licensed under CC BY 4.0. Source: OpenRouter (openrouter.ai/rankings), as of 2026-10-05. Licensed under CC BY 4.0. https://openrouter.ai/rankings