Loading page…
Preferred languages
More languages supported by Google Translate
Open the selector to load the current language catalog.
Translations are provided by Google Translate.
Loading page…
Preferred languages
More languages supported by Google Translate
Open the selector to load the current language catalog.
Translations are provided by Google Translate.
LMArena · Agent · Steering
LMArena's Steering board: models ranked by people voting between two anonymous agent sessions. Every model LMArena ranks is listed; those in the ModelScale catalog link to their page.
Each point is a model with a published LMArena Steering value and complete token prices, on the 75% input / 25% output blend. Frontier models are spaced evenly along the cost axis.
68 models, as LMArena published them on 2026-11-18.
| Rank | Model | Maker | Score | 95% interval | Sessions | Licence |
|---|---|---|---|---|---|---|
| 1 | Anthropic | 0.245 | 0.223–0.267 | 7,564 | Proprietary | |
| 2 | 0.212 | 0.191–0.233 | 3,502 | Proprietary | ||
| 3 | Anthropic | 0.180 | 0.161–0.199 | 7,697 | Proprietary | |
| 4 | 0.179 | 0.173–0.185 | 37,093 | Proprietary | ||
| 5 | 0.174 | 0.158–0.190 | 5,610 | Proprietary | ||
| 6 | Anthropic | 0.148 | 0.122–0.174 | 4,761 | Proprietary | |
| 7 | Anthropic | 0.141 | 0.131–0.150 | 15,464 | Proprietary | |
| 8 | OpenAI | 0.140 | 0.112–0.168 | 2,444 | Proprietary | |
| 9 | OpenAI | 0.139 | 0.121–0.158 | 9,842 | Proprietary | |
| 10 | Anthropic | 0.135 | 0.117–0.154 | 10,074 | Proprietary | |
| 11 | OpenAI | 0.134 | 0.108–0.159 | 2,999 | Proprietary | |
| 12 | OpenAI | 0.132 | 0.120–0.145 | 18,545 | Proprietary | |
| 13 | 0.122 | 0.112–0.132 | 32,023 | Proprietary | ||
| 14 | OpenAI | 0.118 | 0.108–0.128 | 16,453 | Proprietary | |
| 15 | OpenAI | 0.115 | 0.103–0.126 | 20,039 | Proprietary | |
| 16 | OpenAI | 0.099 | 0.093–0.105 | 38,505 | Proprietary | |
| 17 | Anthropic | 0.095 | 0.076–0.114 | 7,150 | Proprietary | |
| 18 | OpenAI | 0.093 | 0.075–0.110 | 8,514 | Proprietary | |
| 19 | Tencent | 0.085 | 0.072–0.098 | 16,752 | Proprietary | |
| 20 | Alibaba | 0.080 | 0.064–0.096 | 8,887 | Proprietary | |
| 21 | Meta | 0.075 | 0.064–0.085 | 21,042 | Proprietary | |
| 22 | Anthropic | 0.055 | 0.040–0.071 | 15,747 | Proprietary | |
| 23 | Meta | 0.048 | 0.036–0.060 | 21,281 | Proprietary | |
| 24 | Moonshot AI | 0.045 | 0.031–0.059 | 15,224 | Kimi K3.1 Pro License | |
| 25 | TMInkling 2 Small | Thinking Machines Lab | 0.044 | 0.031–0.056 | 14,721 | Apache 2.0 |
| 26 | OpenAI | 0.038 | 0.023–0.053 | 16,132 | Proprietary | |
| 27 | Anthropic | 0.024 | 0.008–0.041 | 10,466 | Proprietary | |
| 28 | OpenAI | 0.024 | 0.018–0.030 | 145,516 | Proprietary | |
| 29 | DeepSeek | 0.023 | 0.017–0.030 | 64,690 | MIT | |
| 30 | Xiaomi | 0.016 | 0.004–0.028 | 22,894 | MIT | |
| 31 | Meta | 0.015 | -0.002–0.032 | 10,351 | Proprietary | |
| 32 | Anthropic | 0.012 | -0.005–0.029 | 10,836 | Proprietary | |
| 33 | Meta | 0.008 | -0.003–0.020 | 52,238 | Proprietary | |
| 34 | Anthropic | 0.008 | -0.006–0.021 | 22,595 | Proprietary | |
| 35 | Xiaomi | 0.001 | -0.010–0.013 | 33,297 | MIT | |
| 36 | OpenAI | -0.009 | -0.015–-0.004 | 67,319 | Proprietary | |
| 37 | Tencent | -0.011 | -0.021–-0.001 | 55,196 | Apache 2.0 | |
| 38 | Alibaba | -0.019 | -0.029–-0.009 | 47,428 | Proprietary | |
| 39 | OpenAI | -0.020 | -0.032–-0.009 | 20,966 | Proprietary | |
| 40 | Z.AI | -0.022 | -0.034–-0.009 | 29,541 | MIT | |
| 41 | Moonshot AI | -0.022 | -0.031–-0.013 | 64,513 | Kimi K3 license | |
| 42 | xAI | -0.025 | -0.043–-0.007 | 15,946 | Proprietary | |
| 43 | -0.028 | -0.040–-0.016 | 28,224 | Proprietary | ||
| 44 | StepFun | -0.028 | -0.047–-0.009 | 10,328 | Proprietary | |
| 45 | MiniMax | -0.030 | -0.050–-0.011 | 9,534 | MiniMax Community License | |
| 46 | Z.AI | -0.033 | -0.045–-0.021 | 49,912 | MIT | |
| 47 | OpenAI | -0.036 | -0.045–-0.026 | 34,473 | Proprietary | |
| 48 | Alibaba | -0.038 | -0.050–-0.027 | 61,601 | qwen-community-1.0 | |
| 49 | xAI | -0.043 | -0.063–-0.022 | 11,146 | Proprietary | |
| 50 | Tencent | -0.049 | -0.070–-0.029 | 11,659 | Apache 2.0 | |
| 51 | Z.AI | -0.066 | -0.078–-0.055 | 58,495 | MIT | |
| 52 | Alibaba | -0.078 | -0.093–-0.063 | 36,818 | Apache 2.0 | |
| 53 | Mistral | -0.090 | -0.122–-0.058 | 3,882 | Modified MIT | |
| 54 | -0.091 | -0.106–-0.076 | 33,209 | Proprietary | ||
| 55 | -0.108 | -0.137–-0.079 | 7,520 | Proprietary | ||
| 56 | MiniMax | -0.114 | -0.134–-0.093 | 11,813 | MiniMax Community License | |
| 57 | RABeam 1.0 501B A23B | Reflection AI | -0.120 | -0.159–-0.082 | 8,226 | Apache 2.0 |
| 58 | Alibaba | -0.135 | -0.162–-0.108 | 6,067 | Proprietary | |
| 59 | Alibaba | -0.142 | -0.166–-0.118 | 10,926 | Proprietary | |
| 60 | SYSpark-X2.6 | synlan | -0.145 | -0.162–-0.128 | 10,562 | Proprietary |
| 61 | Meta | -0.147 | -0.163–-0.131 | 32,469 | Proprietary | |
| 62 | Meta | -0.178 | -0.191–-0.164 | 55,532 | Proprietary | |
| 63 | NVIDIA | -0.181 | -0.193–-0.170 | 28,963 | OpenMDW-1.1 | |
| 64 | Mistral | -0.186 | -0.216–-0.156 | 6,118 | Modified MIT | |
| 65 | Upstage | -0.218 | -0.258–-0.177 | 6,336 | Proprietary | |
| 66 | NVIDIA | -0.227 | -0.288–-0.166 | 1,769 | OpenMDW-1.1 | |
| 67 | TMInkling | Thinking Machines Lab | -0.248 | -0.281–-0.214 | 8,530 | Apache 2.0 |
| 68 | TMInkling-Small | Thinking Machines Lab | -0.298 | -0.337–-0.258 | 6,564 | Apache 2.0 |
Source: LMArena, CC BY 4.0. Read once a day. All leaderboards