模型榜单
Claude Opus 5.570.2 分 · ¥53 / 百万 Token最佳性价比最佳编程单项最强
Gemini 4 Argon79 分 · 智能体最佳智能体开放权重最强
DeepSeek-V4.1-Flash-101749.7 分 · 综合第 13 名开放权重最强单项第一
GPT-6 Astra68 分 · 科学推理科学推理第一综合排名
怎么算| 变化 | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | Claude Opus 5.5Anthropic | 70.2 | 77168.4 | 871810 | 71100.0% | 5331.7% | 6373.5% | 6672.2% | 721534 | 751539 | 8/8 | |
| 2 | Gemini 4 ArgonGoogle | 69.7 | 76166.4 | 611679 | 6699.0% | 6527.1% | 7982.4% | 7076.0% | 701527 | 721533 | 8/8 | |
| 3 | Claude Fable 5.1Anthropic | 66.4 | 74166.9 | 711746 | 70100.0% | 6631.1% | 5368.6% | 6570.8% | 661512 | 671519 | 8/8 | |
| 4 | GPT-6 AstraOpenAI | 62.1 | 65169.1 | 581773 | 7399.0% | 6831.7% | 6664.7% | 6975.6% | 391420 | 401454 | 8/8 | |
| 5 | Claude Sonnet 5.5Anthropic | 61.7 | 65166.2 | 681770 | 69100.0% | 6731.4% | 6875.5% | 4146.5% | 541470 | 551492 | 8/8 | |
| 6 | GPT-6.1 TerraOpenAI | 60.8 | 86167.8 | 441673 | 7890.9% | 5928.0% | 5968.3% | 5157.3% | 缺 | 缺 | 6/8 | |
| 7 | GPT-6.1 SolOpenAI | 59.9 | 65168.9 | 541751 | 7399.0% | 6731.7% | 5360.0% | 6873.9% | 401425 | 451467 | 8/8 | |
| 8 |  Muse WonderMeta | 57.0 | 61162.0 | 551652 | 5884.0% | 6028.3% | 5766.1% | 缺 | 591488 | 561494 | 7/8 | |
| 9 |  MiMo-V2.6 ProXiaomi | 51.2 | 571491 | 541625 | 4670.0% | 5626.6% | 4254.4% | 缺 | 521464 | 551492 | 7/8 | |
| 10 | GPT-6 SolOpenAI | 50.7 | ▲1上升 1 名 | 46164.1 | 511678 | 6583.0% | 6330.9% | 5554.3% | 5560.7% | 271378 | 221411 | 8/8 |
| 11 | Muse Spark 1.3Meta | 50.5 | ▼1下降 1 名 | 51158.0 | 541656 | 4458.0% | 5426.0% | 5058.6% | 缺 | 521464 | 501478 | 7/8 |
| 12 | Gemini 3.8 FlashGoogle | 50.1 | 54158.6 | 391583 | 3648.0% | 5118.3% | 4864.3% | 6469.7% | 601491 | 561493 | 8/8 | |
| 13 | DeepSeek-V4.1-Flash-1017DeepSeek | 49.7 | 51159.3 | 461665 | 4371.2% | 5124.0% | 7069.8% | 3742.5% | 缺 | 缺 | 6/8 | |
| 14 |  Kimi K3Moonshot AI | 47.6 | 50158.5 | 541656 | 5187.0% | 5023.4% | 3447.8% | 4550.6% | 501457 | 531485 | 8/8 | |
| 15 |  Qwen3.8 MaxAlibaba | 47.1 | 49157.4 | 401672 | 4458.0% | 4520.0% | 5963.3% | 4045.8% | 561478 | 501480 | 8/8 | |
| 16 |  GLM-5.3Zhipu AI | 43.2 | 46156.9 | 491623 | 3649.0% | 3819.1% | 4656.6% | 3541.0% | 501459 | 511481 | 8/8 | |
| 17 |  Grok 4.7xAI | 36.5 | 30156.1 | 471637 | 1626.0% | 4218.0% | 4954.6% | 5056.0% | 271381 | 201406 | 8/8 | |
| 18 | Muse Spark OpenMeta | 35.4 | 36155.4 | 381526 | 4169.8% | 2614.9% | 533.4% | 5156.6% | 缺 | 缺 | 6/8 | |
| 19 | Claude Haiku 5.5Anthropic | 34.4 | 39160.2 | 361589 | 5489.0% | 3922.6% | 3856.1% | 1823.8% | 141336 | 231415 | 8/8 | |
| 20 | DeepSeek-V4.1-FlashDeepSeek | 34.4 | 44156.5 | 311619 | 3754.0% | 2914.3% | 1639.5% | 3742.9% | 441437 | 481474 | 8/8 | |
| 21 | GPT-6 LunaOpenAI | 32.8 | 30156.7 | 341579 | 5064.0% | 3719.4% | 2444.3% | 3641.4% | 221362 | 201406 | 8/8 | |
| 22 |  Mistral Large 4Mistral | 24.1 | 29152.6 | 301544 | 1610.0% | 1910.6% | 3751.6% | 1520.0% | 221361 | 251419 | 8/8 | |
| 23 |  MiniMax-M3MiniMax | 23.4 | 19147.6 | 81483 | 1518.0% | 203.7% | 1937.7% | 缺 | 331400 | 301431 | 7/8 | |
| — | Gemini 4 CarbonGoogle | 未公开可用，暂不排名 | 751540 | 761768 | 缺 | 7836.0% | 74$16,484 | 缺 | 711531 | 751539 | 6/8 | |
| — | Kimi K3.1 ProMoonshot AI | 新发布，评测待更新 | NEW新上榜 | 缺 | 601695 | 5991.0% | 5124.6% | 缺 | 缺 | 缺 | 缺 | 3/8 |
| — | Muse Wonder FullMeta | 新发布，评测待更新 | 缺 | 641725 | 缺 | 缺 | 缺 | 缺 | 缺 | 缺 | 1/8 | |
| — | MiniMax-M3.1 FlashMiniMax | 新发布，评测待更新 | 缺 | 361582 | 缺 | 缺 | 缺 | 缺 | 缺 | 缺 | 1/8 | |
| — |  Step 5 PreviewStepFun | 评测不足 4 项 | 缺 | 441571 | 2942.0% | 4220.9% | 缺 | 缺 | 缺 | 缺 | 3/8 | |
| — | 编BeamReflection | 新发布，评测待更新 | 321425 | 71436 | 缺 | 缺 | 缺 | 缺 | 301390 | 301431 | 4/8 |
能力雷达
怎么看斩杀线
怎么算价格 × 能力的性价比曲线。性价比 Top 10 都在曲线上或曲线之上；✕ 低于曲线，竖虚线是差的分数。
性价比 Top 10
- 1Claude Opus 5.570.2¥53129%
- 2Gemini 4 Argon69.7¥53128%
- 3DeepSeek-V4.1-Flash-101749.7¥3.5124%
- 4Claude Sonnet 5.561.7¥27120%
- 5GPT-6.1 Sol59.9¥27116%
- 6Claude Fable 5.166.4¥133116%
- 7GPT-6 Astra62.1¥133109%
- 8Claude Haiku 5.534.4¥1.3105%
- 9Gemini 3.8 Flash50.1¥20100%
- 10GPT-6 Luna32.8¥1.3100%