OnlySOTA

汇集前沿人工智能大模型的最新排行及斩杀线

两份榜单一平均,读起来就成了一份意见一致的榜。此处从不平均——分歧才是要读的东西。

排行 · AA 38–63 · ARENA 1363–1513
  1. 1763.11493
  2. 1262.11508
  3. 31660.91482
  4. 44660.91461
  5. 51059.71489
  6. 613新上榜59.51487
  7. 758.1
  8. 857.7
  9. 4956.81498
  10. 104456.61465
  11. 91156.01490
  12. 123555.81470
  13. 134555.31461
  14. 145253.21458
  15. 153452.61470
  16. 166352.31451
  17. 17新上榜52.0
  18. 1851.8
  19. 192051.61481
  20. 142047.71486
  21. 2147.4
  22. 2122估算46.71478
  23. 23估算45.5
  24. 247645.41443
  25. 25估算45.3
  26. 2645.3
  27. 2743.0
  28. 284042.91468
  29. 298042.31440
  30. 305642.21457
  31. 3141.7
  32. 3217241.61376
  33. 3313641.21405
  34. 34估算39.9
  35. 3539.7
  36. 365339.41457
  37. 3710238.31426
  38. 389138.01434
  39. 39斩杀者37.8
查看更多 · AA 榜共 192 条 →
斩杀线 · 指数 × 价格
ARTIFICIAL ANALYSIS INTELLIGENCE INDEX · 线性刻度(从 0 开始)2040600$0.05$0.25$1.00$5.00$20.00斩杀者Ling 3.0 Flash37.8 · $0.11 63.1$10.00Claude Opus 5SOTA低成本被斩杀BLENDED PRICE, USD PER 1M TOKENS (3:1 INPUT:OUTPUT) · LOG10
斩杀线SOTA其余 · 颜色表示分区空心 · 该源估算

Claude Opus 5

anthropic/claude-opus-5 · 闭源权重 · $10.00 · SOTA

6数据源
17
14配置
各榜排名 · 6 数据源
Artificial Analysis LLM Leaderboard63.1
1Arena — Agent0.125
Artificial Analysis Coding Agents0.667
1Arena — Code1691
7Arena — Text1493
9Arena — Vision1292

每个条形都画在该数据源自己拟合出的范围上,位置在行与行之间不可比。名次是该源自己的名次;该源不发布名次的,这里既没有数字,也没有条形。

推理强度档位 · 14 配置
Artificial Analysis LLM Leaderboardmax 最优
minlowmedhighxhimax52.558.661.562.563.1
Arena — Agent · unknownhigh 最优
minlowmedhighxhimax0.1250.120
Artificial Analysis Coding Agents · Claude Codexhi 最优
minlowmedhighxhimax0.5680.6190.6340.6670.655
Arena — Codemax 最优
minlowmedhighxhimax16631691
Arena — Texthigh 最优
minlowmedhighxhimax14931487
Arena — Vision仅跑了 high
minlowmedhighxhimax1292

Six named efforts, same columns on every track; a greyed name is a configuration this source did not publish. Scores are each source's own and are not compared across tracks.