排行榜
AI 模型排行榜(2026年11月):按得分、价格与速度评出的最佳大模型
哪个模型最好,取决于你的需求。其中三个榜单来自 LMArena 的社区排名;另外两个只有 WayJet 能测:性价比,以及真实流量下的速度。可以在 WayJet 上调用的模型旁边有一个按钮。 榜单是怎么做的
综合对话
按 LMArena 的文本得分(含风格控制)排名:用户比较两个匿名回答,投票选出更好的一个。
来源: LMArena (CC BY 4.0),数据截至 2026-11-09.
| 排名 | 模型 | 得分 | WayJet 上 |
|---|---|---|---|
| 1 | Gemini 4 Carbon (High)google | 1526 | |
| 2 | Gemini 4 Argon (High)google | 1522 | |
| 3 | Muse Wonder (Max)meta | 1511 | |
| 4 | Claude Opus 5.5 (High)anthropic | 1506 | 试用 |
| 5 | Claude Fable 5 (High)anthropic | 1504 | 试用 |
| 6 | Claude Opus 4.6 (High)anthropic | 1503 | 试用 |
| 7 | Claude Fable 5.1 (Max)anthropic | 1503 | 试用 |
| 8 | Claude Opus 4.7 (High)anthropic | 1501 | 试用 |
| 9 | Gemini 3.8 Flash (High)google | 1498 | 试用 |
| 10 | Claude Opus 4.6anthropic | 1497 | 试用 |
| 11 | Muse Spark 1.3 (Max)meta | 1496 | |
| 12 | Claude Opus 4.7anthropic | 1494 | 试用 |
| 13 | Muse Spark 1.2 (xHigh)meta | 1494 | |
| 14 | Muse Spark 1.1meta | 1491 | |
| 15 | Claude Opus 5 (High)anthropic | 1490 | 试用 |
| 16 | Gemini 3.7 Flash (High)google | 1489 | |
| 17 | Muse Sparkmeta | 1489 | |
| 18 | Claude Opus 5 (Max)anthropic | 1489 | 试用 |
| 19 | Kimi K3 (Max)moonshot | 1487 | 试用 |
| 20 | Gemini 3.1 Pro Previewgoogle | 1487 | |
| 21 | Gemini 3 Progoogle | 1486 | |
| 22 | GPT-5.6 Sol (xHigh)openai | 1485 | 试用 |
| 23 | Qwen3.8-Maxalibaba | 1484 | 试用 |
| 24 | Kimi K3.1 Pro (Max)moonshot | 1483 | 试用 |
| 25 | Gemini 3.6 Flash (High)google | 1483 | 试用 |
| 26 | GLM-5.3 (Max)zai | 1483 | |
| 27 | GPT-6.1 Sol (Max)openai | 1482 | 试用 |
| 28 | Claude Opus 4.8 (High)anthropic | 1482 | 试用 |
| 29 | GPT-5.5 (High)openai | 1481 | |
| 30 | MiMo-V2.6-Proxiaomi | 1480 | |
| 31 | Gemini 3.5 Flash (High)google | 1478 | |
| 32 | Gemini 3.5 Flash (Medium)google | 1476 | |
| 33 | GPT-5.5openai | 1476 | |
| 34 | GPT-5.2 Chat (20260210)openai | 1476 | 试用 |
| 35 | GPT-5.4 (High)openai | 1475 | 试用 |
| 36 | GLM-5.2 (Max)zai | 1475 | |
| 37 | DeepSeek-V4.1-Flash-20260910 (Max)deepseek | 1475 | |
| 38 | Grok 4.20 Beta 1xai | 1475 | |
| 39 | Claude Opus 4.8anthropic | 1474 | 试用 |
| 40 | GLM-5.3-Flashzai | 1474 | |
| 41 | GPT-6 Astra (Max)openai | 1474 | 试用 |
| 42 | Qwen3.7-Max-Previewalibaba | 1474 | |
| 43 | Claude Opus 4.5 (High, 32K, 20251101)anthropic | 1474 | |
| 44 | GPT-5.5 Instantopenai | 1473 | |
| 45 | Gemini 3 Flashgoogle | 1473 | |
| 46 | Claude Sonnet 5.5 (xHigh)anthropic | 1473 | 试用 |
| 47 | Grok 4.20 Beta (Reasoning, 0309)xai | 1472 | |
| 48 | Claude Sonnet 4.6anthropic | 1472 | 试用 |
| 49 | Grok 4.20 Multi-Agent Beta (0309)xai | 1471 | |
| 50 | Claude Opus 4.5 (20251101)anthropic | 1470 | 试用 |
编程
按 LMArena 的 WebDev 得分排名:模型根据提示做出一个能运行的网页应用,由用户投票评判结果。
来源: LMArena (CC BY 4.0),数据截至 2026-11-09.
| 排名 | 模型 | 得分 | WayJet 上 |
|---|---|---|---|
| 1 | Claude Opus 5.5 (Max)anthropic | 1818 | 试用 |
| 2 | Claude Sonnet 5.5 (xHigh)anthropic | 1774 | 试用 |
| 3 | Gemini 4 Carbon (High)google | 1768 | |
| 4 | GPT-6 Astra (Max)openai | 1767 | 试用 |
| 5 | Claude Fable 5.1 (Max)anthropic | 1745 | 试用 |
| 6 | GPT-6.1 Sol (Max)openai | 1743 | 试用 |
| 7 | Muse Wonder Full (Max)meta | 1724 | |
| 8 | Claude Sonnet 5.5 (High)anthropic | 1713 | 试用 |
| 9 | Claude Opus 5 (Max)anthropic | 1691 | 试用 |
| 10 | Kimi K3.1 Pro (Max)moonshot | 1688 | 试用 |
| 11 | Gemini 4 Argon (High)google | 1683 | |
| 12 | GPT-6.1 Terra (Max)openai | 1679 | 试用 |
| 13 | Qwen3.8-Max (0902)alibaba | 1677 | 试用 |
| 14 | GPT-6 Sol (Max)openai | 1674 | 试用 |
| 15 | Qwen3.8-Maxalibaba | 1670 | 试用 |
| 16 | DeepSeek-V4.1-Flash-20261017 (Max)deepseek | 1664 | |
| 17 | Kimi K3 (Max)moonshot | 1657 | 试用 |
| 18 | Claude Opus 5 (High)anthropic | 1657 | 试用 |
| 19 | Muse Spark 1.3 (Max)meta | 1654 | |
| 20 | Muse Wonder (Max)meta | 1650 | |
| 21 | Inkling 2 Smallthinky | 1650 | |
| 22 | Grok 4.7 (xHigh)xai | 1639 | |
| 23 | MiMo-V2.6-Flashxiaomi | 1637 | |
| 24 | Hy4 previewtencent | 1632 | |
| 25 | Claude Fable 5 (High)anthropic | 1625 | 试用 |
| 26 | Qwen3.8-Flash-Nextalibaba | 1625 | |
| 27 | MiMo-V2.6-Proxiaomi | 1624 | |
| 28 | GLM-5.3 (Max)zai | 1622 | |
| 29 | DeepSeek-V4.1-Flash-20260910 (Max)deepseek | 1621 | |
| 30 | Muse Spark 1.3 (xHigh)meta | 1621 | |
| 31 | Grok 4.6 (High)xai | 1618 | |
| 32 | GPT-5.6 Sol (xHigh, Codex Harness)openai | 1615 | 试用 |
| 33 | GLM-5.3-Flashzai | 1609 | |
| 34 | Muse Ember (Max)meta | 1608 | |
| 35 | GLM-5.2 (Max)zai | 1608 | |
| 36 | Claude Haiku 5.5 (High)anthropic | 1595 | 试用 |
| 37 | Qwen3.8-27Balibaba | 1591 | 试用 |
| 38 | Gemini 3.7 Flash (High)google | 1588 | |
| 39 | Gemini 3.8 Flash (High)google | 1584 | 试用 |
| 40 | DeepSeek-V4-Flash (High)deepseek | 1582 | 试用 |
| 41 | MiniMax-M3.1-Flashminimax | 1582 | 试用 |
| 42 | DeepSeek-V4-Pro (High, 20260813)deepseek | 1581 | 试用 |
| 43 | GPT-6 Luna (Max)openai | 1578 | 试用 |
| 44 | Step 5 Preview (High)stepfun | 1573 | |
| 45 | Mistral Large 4mistral | 1562 | |
| 46 | Claude Opus 4.7 (High)anthropic | 1557 | 试用 |
| 47 | Claude Opus 4.8 (High)anthropic | 1555 | 试用 |
| 48 | Claude Opus 4.7anthropic | 1555 | 试用 |
| 49 | Grok 4.5xai | 1552 | |
| 50 | Claude Opus 4.6 (High)anthropic | 1546 | 试用 |
视觉与多模态
按 LMArena 的视觉得分(含风格控制)排名:带有图片的问题。
来源: LMArena (CC BY 4.0),数据截至 2026-11-09.
| 排名 | 模型 | 得分 | WayJet 上 |
|---|---|---|---|
| 1 | Claude Fable 5 (High)anthropic | 1309 | 试用 |
| 2 | Claude Opus 4.7anthropic | 1300 | 试用 |
| 3 | Qwen3.8-Maxalibaba | 1299 | 试用 |
| 4 | Claude Opus 4.6 (High)anthropic | 1299 | 试用 |
| 5 | Claude Opus 4.7 (High)anthropic | 1298 | 试用 |
| 6 | Claude Opus 5.5 (High)anthropic | 1295 | 试用 |
| 7 | Muse Sparkmeta | 1294 | |
| 8 | Gemini 3.7 Flash (High)google | 1294 | |
| 9 | Muse Spark 1.2 (xHigh)meta | 1294 | |
| 10 | Claude Opus 4.6anthropic | 1293 | 试用 |
| 11 | Gemini 3.8 Flash (High)google | 1293 | 试用 |
| 12 | GPT-6.1 Sol (Max)openai | 1293 | 试用 |
| 13 | Muse Spark 1.3 (Max)meta | 1292 | |
| 14 | Gemini 3 Progoogle | 1289 | |
| 15 | Claude Fable 5.1 (Max)anthropic | 1289 | 试用 |
| 16 | GPT-5.6 Sol (xHigh)openai | 1288 | 试用 |
| 17 | Muse Spark 1.3 (xHigh)meta | 1288 | |
| 18 | Claude Opus 4.8 (High)anthropic | 1286 | 试用 |
| 19 | Claude Opus 5 (High)anthropic | 1285 | 试用 |
| 20 | GPT-5.5openai | 1285 | |
| 21 | GPT-6 Astra (Max)openai | 1283 | 试用 |
| 22 | GPT-5.5 (High)openai | 1283 | |
| 23 | GPT-5.4 (High)openai | 1282 | 试用 |
| 24 | Gemini 3.5 Flash (Medium)google | 1282 | |
| 25 | Gemini 3.5 Flash (High)google | 1281 | |
| 26 | Grok 4.5xai | 1280 | |
| 27 | Gemini 3.1 Pro Previewgoogle | 1280 | |
| 28 | Gemini 3.6 Flash (High)google | 1280 | 试用 |
| 29 | Muse Spark 1.1meta | 1279 | |
| 30 | Claude Opus 4.8anthropic | 1279 | 试用 |
| 31 | GPT-5.4openai | 1278 | 试用 |
| 32 | GPT-5.2 Chat (20260210)openai | 1278 | 试用 |
| 33 | GPT-5.5 Instantopenai | 1277 | |
| 34 | Claude Sonnet 4.6anthropic | 1275 | 试用 |
| 35 | Gemini 3 Flashgoogle | 1272 | |
| 36 | GLM-5.3-Flashzai | 1272 | |
| 37 | GPT-5.6 Terra (xHigh)openai | 1266 | 试用 |
| 38 | Gemini 3.5 Flash-Litegoogle | 1266 | |
| 39 | Gemma 4 31Bgoogle | 1265 | |
| 40 | Kimi K2.6moonshot | 1265 | 试用 |
| 41 | Grok 4.6 (High)xai | 1263 | |
| 42 | DeepSeek-V4.1-Flash-20260910 (Max)deepseek | 1263 | |
| 43 | Claude Sonnet 5.5 (xHigh)anthropic | 1263 | 试用 |
| 44 | Qwen3.7-Plusalibaba | 1262 | 试用 |
| 45 | Claude Sonnet 5 (High)anthropic | 1261 | 试用 |
| 46 | Gemini 3 Flash (Minimal)google | 1260 | |
| 47 | GPT-6 Sol (Max)openai | 1258 | 试用 |
| 48 | Dola Seed 2.0 Probytedance | 1256 | |
| 49 | Grok 4.20 Beta (Reasoning, 0309)xai | 1256 | |
| 50 | GPT-5.6 Luna (xHigh)openai | 1255 | 试用 |
性价比
综合得分除以该模型在 WayJet 上每百万 token 的价格(输入与输出按 3:1 加权)。仅包含可在 WayJet 上调用的模型。
来源: LMArena (CC BY 4.0),数据截至 2026-11-09;价格:WayJet,以本页生成当天为准.
| 排名 | 模型 | 每百万 token 价格 | 每美元得分 | WayJet 上 |
|---|---|---|---|---|
| 1 | google.gemma-3-4b-itgoogle | $0.05 | 26,065 | 试用 |
| 2 | gpt-5-nanoopenai | $0.14 | 9,728 | 试用 |
| 3 | google.gemma-3-12b-itgoogle | $0.14 | 9,584 | 试用 |
| 4 | qwen3-30b-a3b-instruct-2507alibaba | $0.15 | 9,277 | 试用 |
| 5 | zai.glm-4.7-flashzai | $0.15 | 8,947 | 试用 |
| 6 | qwen3.5-flashalibaba | $0.17 | 7,977 | 试用 |
| 7 | gpt-4.1-nanoopenai | $0.17 | 7,556 | 试用 |
| 8 | gpt-6-lunaopenai | $0.20 | 7,204 | 试用 |
| 9 | claude-haiku-5-5anthropic | $0.20 | 7,112 | 试用 |
| 10 | deepseek-v4-flashdeepseek | $0.26 | 5,470 | 试用 |
| 11 | google.gemma-3-27b-itgoogle | $0.27 | 5,104 | 试用 |
| 12 | gpt-4o-miniopenai | $0.26 | 5,020 | 试用 |
| 13 | qwen.qwen3-next-80b-a3balibaba | $0.41 | 3,391 | 试用 |
| 14 | gpt-5.6-lunaopenai | $0.45 | 3,229 | 试用 |
| 15 | qwen3-235b-a22b-instruct-2507alibaba | $0.50 | 2,826 | 试用 |
| 16 | MiniMax-M3minimax | $0.53 | 2,735 | 试用 |
| 17 | MiniMax-M2.7minimax | $0.53 | 2,695 | 试用 |
| 18 | MiniMax-M2.5minimax | $0.53 | 2,649 | 试用 |
| 19 | MiniMax-M2minimax | $0.53 | 2,559 | 试用 |
| 20 | qwen3.7-plusalibaba | $0.70 | 2,079 | 试用 |
| 21 | qwen3.5-35b-a3balibaba | $0.69 | 2,028 | 试用 |
| 22 | gpt-5-miniopenai | $0.69 | 2,022 | 试用 |
| 23 | gpt-4.1-miniopenai | $0.70 | 1,976 | 试用 |
| 24 | qwen3.8-27balibaba | $0.78 | 1,850 | 试用 |
| 25 | qwen3.5-27balibaba | $0.82 | 1,707 | 试用 |
| 26 | gemini-2.5-flashgoogle | $0.85 | 1,658 | 试用 |
| 27 | gpt-3.5-turboopenai | $0.75 | 1,634 | 试用 |
| 28 | deepseek.v3.2deepseek | $0.93 | 1,536 | 试用 |
| 29 | deepseek-v4-prodeepseek | $0.99 | 1,479 | 试用 |
| 30 | zai.glm-4.7zai | $1.00 | 1,441 | 试用 |
| 31 | qwen.qwen3-vl-235b-a22balibaba | $1.06 | 1,330 | 试用 |
| 32 | qwen3.5-122b-a10balibaba | $1.10 | 1,288 | 试用 |
| 33 | qwen3.6-plusalibaba | $1.13 | 1,283 | 试用 |
| 34 | moonshotai.kimi-k2.5moonshot | $1.20 | 1,208 | 试用 |
| 35 | qwen3.5-397b-a17balibaba | $1.35 | 1,068 | 试用 |
| 36 | gemini-3.8-flashgoogle | $1.50 | 998 | 试用 |
| 37 | gemini-3.6-flashgoogle | $1.50 | 988 | 试用 |
| 38 | zai.glm-5zai | $1.55 | 940 | 试用 |
| 39 | kimi-k2.6moonshot | $1.71 | 853 | 试用 |
| 40 | qwen3-next-80b-a3b-thinkingalibaba | $1.88 | 730 | 试用 |
| 41 | claude-haiku-4-5-20251001anthropic | $2.00 | 707 | 试用 |
| 42 | qwen3-maxalibaba | $2.40 | 593 | 试用 |
| 43 | qwen3.6-max-previewalibaba | $2.92 | 499 | 试用 |
| 44 | qwen3.8-maxalibaba | $3.00 | 495 | 试用 |
| 45 | qwen-maxalibaba | $2.80 | 471 | 试用 |
| 46 | qwen3-coder-480b-a35b-instructalibaba | $3.00 | 463 | 试用 |
| 47 | gpt-5.1openai | $3.44 | 423 | 试用 |
| 48 | gpt-5openai | $3.44 | 417 | 试用 |
| 49 | gpt-4.1openai | $3.50 | 404 | 试用 |
| 50 | kimi-k3.1-promoonshot | $4.00 | 371 | 试用 |

