AI 文档理解模型排行榜

查看 Arena 文档任务榜的评分。

评测来源 Arena · 文档原榜发布 查看原榜评分方法

Arena Score

44 个模型
Arena · 文档 · Arena Score
排名模型 / 开发机构Arena Score投票95% CI许可详情
01claude-fable-5Anthropic1499Arena Score投票849795% CI1491–1507许可Proprietary投票849795% CI1491–1507许可Proprietary详情
02claude-opus-4-6-highAnthropic1495Arena Score投票2792995% CI1489–1502许可Proprietary投票2792995% CI1489–1502许可Proprietary详情
03claude-opus-4-7Anthropic1494Arena Score投票2219295% CI1488–1501许可Proprietary投票2219295% CI1488–1501许可Proprietary详情
04claude-opus-4-6Anthropic1494Arena Score投票4126095% CI1488–1500许可Proprietary投票4126095% CI1488–1500许可Proprietary详情
05claude-opus-4-7-highAnthropic1493Arena Score投票2195795% CI1486–1500许可Proprietary投票2195795% CI1486–1500许可Proprietary详情
06claude-fable-5.1-maxAnthropic1493Arena Score投票140395% CI1478–1508许可Proprietary投票140395% CI1478–1508许可Proprietary详情
07claude-opus-5-highAnthropic1490Arena Score投票879495% CI1483–1497许可Proprietary投票879495% CI1483–1497许可Proprietary详情
08gpt-5.6-sol-xhighOpenAI1489Arena Score投票481895% CI1480–1498许可Proprietary投票481895% CI1480–1498许可Proprietary详情
09claude-opus-4-8-highAnthropic1486Arena Score投票1155195% CI1479–1493许可Proprietary投票1155195% CI1479–1493许可Proprietary详情
10gpt-5.5OpenAI1483Arena Score投票2127995% CI1476–1489许可Proprietary投票2127995% CI1476–1489许可Proprietary详情
11gpt-5.5-highOpenAI1480Arena Score投票2094495% CI1474–1487许可Proprietary投票2094495% CI1474–1487许可Proprietary详情
12muse-spark-1.1Meta1479Arena Score投票488595% CI1471–1488许可Proprietary投票488595% CI1471–1488许可Proprietary详情
13claude-sonnet-4-6Anthropic1478Arena Score投票5781395% CI1472–1484许可Proprietary投票5781395% CI1472–1484许可Proprietary详情
14gpt-5.6-terra-xhighOpenAI1477Arena Score投票578795% CI1468–1485许可Proprietary投票578795% CI1468–1485许可Proprietary详情
15claude-sonnet-5-highAnthropic1476Arena Score投票741195% CI1468–1483许可Proprietary投票741195% CI1468–1483许可Proprietary详情
16claude-opus-4-8Anthropic1475Arena Score投票1212895% CI1468–1482许可Proprietary投票1212895% CI1468–1482许可Proprietary详情
17gpt-6-astra-maxOpenAI1473Arena Score投票162195% CI1458–1487许可Proprietary投票162195% CI1458–1487许可Proprietary详情
18gpt-5.4OpenAI1470Arena Score投票3333195% CI1464–1476许可Proprietary投票3333195% CI1464–1476许可Proprietary详情
19muse-spark-1.3-maxMeta1468Arena Score投票100695% CI1450–1486许可Proprietary投票100695% CI1450–1486许可Proprietary详情
20muse-sparkMeta1467Arena Score投票108495% CI1449–1486许可Proprietary投票108495% CI1449–1486许可Proprietary详情
21claude-opus-4-5-20251101Anthropic1465Arena Score投票798195% CI1454–1475许可Proprietary投票798195% CI1454–1475许可Proprietary详情
22grok-4.5xAI1464Arena Score投票496595% CI1456–1473许可Proprietary投票496595% CI1456–1473许可Proprietary详情
23gpt-5.6-luna-xhighOpenAI1462Arena Score投票576695% CI1454–1470许可Proprietary投票576695% CI1454–1470许可Proprietary详情
24grok-4.6-highxAI1461Arena Score投票225895% CI1449–1473许可Proprietary投票225895% CI1449–1473许可Proprietary详情
25gemini-3.5-flash-mediumGoogle1461Arena Score投票650095% CI1453–1468许可Proprietary投票650095% CI1453–1468许可Proprietary详情
26gpt-5.5-instantOpenAI1460Arena Score投票849795% CI1452–1469许可Proprietary投票849795% CI1452–1469许可Proprietary详情
27gemini-3.1-pro-previewGoogle1459Arena Score投票4945795% CI1454–1464许可Proprietary投票4945795% CI1454–1464许可Proprietary详情
28gemini-3.6-flash-highGoogle1453Arena Score投票297595% CI1443–1464许可Proprietary投票297595% CI1443–1464许可Proprietary详情
29gemini-3.5-flash-highGoogle1453Arena Score投票406195% CI1444–1463许可Proprietary投票406195% CI1444–1463许可Proprietary详情
30qwen3.7-plusAlibaba1453Arena Score投票459195% CI1443–1462许可Proprietary投票459195% CI1443–1462许可Proprietary详情
31gemini-3-proGoogle1451Arena Score投票1076995% CI1442–1460许可Proprietary投票1076995% CI1442–1460许可Proprietary详情
32kimi-k2.6Moonshot AI1450Arena Score投票1129195% CI1442–1458许可Modified MIT投票1129195% CI1442–1458许可Modified MIT详情
33claude-sonnet-4-5-20250929Anthropic1449Arena Score投票3145595% CI1443–1455许可Proprietary投票3145595% CI1443–1455许可Proprietary详情
34gemma-4-31bGoogle1445Arena Score投票1232695% CI1437–1452许可Apache 2.0投票1232695% CI1437–1452许可Apache 2.0详情
35grok-4.20-beta-0309-reasoningxAI1439Arena Score投票2094895% CI1432–1447许可Proprietary投票2094895% CI1432–1447许可Proprietary详情
36gemini-3-flashGoogle1439Arena Score投票715895% CI1430–1449许可Proprietary投票715895% CI1430–1449许可Proprietary详情
37kimi-k2.5-thinkingMoonshot AI1436Arena Score投票2156995% CI1429–1443许可Modified MIT投票2156995% CI1429–1443许可Modified MIT详情
38minimax-m3MiniMax1433Arena Score投票631495% CI1425–1441许可MiniMax Community License投票631495% CI1425–1441许可MiniMax Community License详情
39gemini-2.5-proGoogle1431Arena Score投票2511095% CI1424–1437许可Proprietary投票2511095% CI1424–1437许可Proprietary详情
40gpt-5.2OpenAI1425Arena Score投票2821295% CI1419–1431许可Proprietary投票2821295% CI1419–1431许可Proprietary详情
41gpt-5.2-highOpenAI1423Arena Score投票710895% CI1414–1433许可Proprietary投票710895% CI1414–1433许可Proprietary详情
42claude-haiku-4-5-20251001Anthropic1419Arena Score投票3467795% CI1413–1425许可Proprietary投票3467795% CI1413–1425许可Proprietary详情
43gpt-5.1OpenAI1416Arena Score投票824495% CI1407–1425许可Proprietary投票824495% CI1407–1425许可Proprietary详情
44glm-5v-turboZ.ai1400Arena Score投票699595% CI1392–1408许可Proprietary投票699595% CI1392–1408许可Proprietary详情

数据来自 Arena 官方公开数据集(CC BY 4.0);保留原榜名次与评分。 官方数据集

未公布的数据记为 —