---
title: "Agent Arena: model rankings and source evidence · Undominated.ai"
canonical: https://undominated.ai/rankings/arena-agent/
description: "Agent Arena ranked by IPS score. Inspect exact model variants, source dates, measurement limits and downloadable evidence."
---

# Agent Arena: model rankings and source evidence · Undominated.ai

> Agent Arena ranked by IPS score. Inspect exact model variants, source dates, measurement limits and downloadable evidence.

[All rankings](/rankings/) / Coding and agents

# Agent Arena

Agent Arena signal scores from the official dataset. Each signal is a separate measurement; it is not a success percentage.

**58 entries** · IPS score · Higher is better

Source publication: 2026-11-04. Retrieved 2026-11-05.

Source: [Arena (lmarena-ai)](https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset) · [CC-BY-4.0](https://creativecommons.org/licenses/by/4.0/).

Ranks use confidence bounds. Several models can share a rank; the rows within a shared rank are displayed by score.

Agent Arena · IPS score

| Significance rank | Model / evaluated variant | IPS score | 95% confidence interval | Observations |
| --- | --- | --- | --- | --- |
| 1 | Claude Opus 5.5 (Max) anthropic | 0.201448 | 0.18059 – 0.222307 | 2,350,984 |
| 2 | Gemini 4 Carbon (High) google | 0.167166 | 0.1595 – 0.174832 | 7,450,409 |
| 1 | Claude Sonnet 5.5 (Max) anthropic | 0.154237 | 0.126475 – 0.181998 | 1,201,743 |
| 2 | GPT-6 Astra (Max) openai | 0.14697 | 0.126318 – 0.167622 | 1,293,489 |
| 3 | Claude Fable 5.1 (Max) anthropic | 0.13857 | 0.119835 – 0.157306 | 2,422,643 |
| 3 | GPT-6.1 Sol (Max) openai | 0.125139 | 0.114051 – 0.136228 | 2,909,711 |
| 3 | GPT-6.1 Terra (Max) openai | 0.11396 | 0.099983 – 0.127938 | 6,713,903 |
| 7 | Gemini 4 Argon (High) google | 0.105252 | 0.096802 – 0.113702 | 6,589,073 |
| 7 | Muse Wonder Full (Max) meta | 0.094226 | 0.084779 – 0.103674 | 5,846,669 |
| 9 | Claude Fable 5 (High) anthropic | 0.078225 | 0.067081 – 0.089369 | 3,486,141 |
| 9 | Claude Opus 5 (Max) anthropic | 0.07662 | 0.06418 – 0.089059 | 4,630,296 |
| 10 | GPT-6 Sol (Max) openai | 0.067781 | 0.052345 – 0.083218 | 3,117,453 |
| 10 | Hy4 tencent | 0.053922 | 0.036817 – 0.071028 | 1,473,452 |
| 11 | Kimi K3.1 Pro (Max) moonshot | 0.052922 | 0.040115 – 0.065729 | 5,477,867 |
| 12 | GPT-5.6 Sol (xHigh) openai | 0.045723 | 0.035355 – 0.05609 | 6,530,847 |
| 15 | MiMo-V2.6-Pro xiaomi | 0.02681 | 0.017665 – 0.035955 | 5,417,543 |
| 13 | Grok 4.7 (xHigh) xai | 0.026802 | 0.012199 – 0.041404 | 3,666,622 |
| 16 | Gemini 3.8 Flash (High) google | 0.02582 | 0.017531 – 0.034108 | 8,498,907 |
| 16 | Muse Spark 1.3 (Max) meta | 0.025114 | 0.016843 – 0.033384 | 8,877,602 |
| 15 | Claude Sonnet 5 (High) anthropic | 0.02443 | 0.012757 – 0.036102 | 5,157,478 |
| 14 | Muse Wonder (Max) meta | 0.024336 | 0.010837 – 0.037834 | 2,088,667 |
| 16 | DeepSeek-V4.1-Flash-20261017 (Max) deepseek | 0.0237 | 0.018043 – 0.029358 | 20,237,992 |
| 16 | GPT-6 Luna (Max) openai | 0.018937 | 0.014526 – 0.023349 | 41,141,530 |
| 16 | GLM-5.3 (Max) zai | 0.010891 | 0.002935 – 0.018847 | 11,340,717 |
| 21 | Kimi K3 (Max) moonshot | 0.008419 | 0.002635 – 0.014204 | 16,147,659 |
| 21 | Hy4 preview tencent | 0.00749 | 0.000546 – 0.014435 | 11,647,463 |
| 21 | GLM-5.2 (Max) zai | 0.006386 | -0.001348 – 0.01412 | 9,213,548 |
| 20 | Claude Haiku 5.5 (High) anthropic | 0.006244 | -0.004024 – 0.016512 | 4,139,248 |
| 23 | Qwen3.8-Max alibaba | 0.004788 | -0.001763 – 0.01134 | 8,376,850 |
| 20 | Grok 4.6 (xHigh) xai | 0.004543 | -0.007223 – 0.016308 | 3,777,170 |
| 21 | MiMo-V2.6-Flash xiaomi | 0.002001 | -0.009278 – 0.01328 | 3,647,059 |
| 21 | DeepSeek-V4-Pro (High, 20260813) deepseek | -0.003004 | -0.020015 – 0.014007 | 932,407 |
| 31 | GPT-5.6 Terra (xHigh) openai | -0.015047 | -0.022869 – -0.007224 | 3,770,129 |
| 31 | Gemini 3.7 Flash (High) google | -0.016025 | -0.024778 – -0.007272 | 6,216,788 |
| 32 | GPT-6 Luna (Medium) openai | -0.018432 | -0.023735 – -0.01313 | 8,786,320 |
| 32 | GLM-5.3-Flash zai | -0.02394 | -0.030866 – -0.017013 | 13,609,316 |
| 32 | MiniMax-M3.1-Flash minimax | -0.028435 | -0.043845 – -0.013025 | 2,553,700 |
| 32 | Step 5 Preview stepfun | -0.028761 | -0.042746 – -0.014777 | 2,816,931 |
| 37 | Qwen3.8-27B alibaba | -0.042751 | -0.051339 – -0.034164 | 8,425,447 |
| 37 | Qwen3.8-Flash-Next alibaba | -0.044295 | -0.051617 – -0.036973 | 16,929,386 |
| 36 | Gemini 3.1 Pro Preview google | -0.04441 | -0.058552 – -0.030268 | 2,970,564 |
| 36 | Beam 1.0 501B A23B reflection | -0.053809 | -0.078863 – -0.028755 | 701,442 |
| 37 | Mistral Large 4 mistral | -0.054627 | -0.073912 – -0.035342 | 447,184 |
| 41 | GPT-6 Luna (Low) openai | -0.059289 | -0.066333 – -0.052244 | 3,349,837 |
| 41 | Qwen3.7-Max alibaba | -0.0652 | -0.077326 – -0.053073 | 2,294,942 |
| 42 | Qwen3.7-Plus alibaba | -0.074599 | -0.090158 – -0.059041 | 1,280,054 |
| 43 | Muse Spark 1.1 meta | -0.076595 | -0.083205 – -0.069984 | 8,564,936 |
| 43 | Muse Spark 1.2 (xHigh) meta | -0.07699 | -0.085443 – -0.068537 | 3,876,357 |
| 42 | Hy3 tencent | -0.077197 | -0.09166 – -0.062734 | 1,234,715 |
| 46 | MiMo-V2.5-Pro xiaomi | -0.09027 | -0.100443 – -0.080098 | 2,517,888 |
| 47 | Spark-X2.6 synlan | -0.099026 | -0.112846 – -0.085207 | 1,027,728 |
| 50 | MiniMax-M3 minimax | -0.103967 | -0.115854 – -0.09208 | 3,672,031 |
| 53 | Mistral Medium 3.5 mistral | -0.13344 | -0.150359 – -0.116521 | 680,776 |
| 53 | Nemotron 3 Ultra nvidia | -0.155971 | -0.188962 – -0.12298 | 228,023 |
| 54 | NVIDIA-Nemotron-3.5-Super-VL-120B-A12B (NVFP4) nvidia | -0.159321 | -0.168096 – -0.150546 | 1,618,598 |
| 53 | Inkling thinky | -0.163321 | -0.17823 – -0.148413 | 1,748,439 |
| 54 | Inkling-Small thinky | -0.170396 | -0.189101 – -0.151692 | 564,788 |
| 58 | Solar Pro 4 upstage | -0.218047 | -0.239465 – -0.196628 | 546,591 |

## What this ranking can tell you

 - The population is the models in this source board, including variants and models without an available API price.
- Ranks use the published confidence bounds: one plus the number of models whose lower bound exceeds this model’s upper bound. Shared ranks are not proof of equal performance.

## Check the source

The original evidence is retained byte for byte. The downloadable board includes the source URLs, revision and SHA-256 hashes.

[Download this board and provenance →](/data/rankings/arena-agent.json)

 Source files and licences - [metadata · a7a159f557a1](/data/rankings/evidence/a7a159f557a1e594a5d41218fe8d53c63c63daa39b655e51d6fc5b482a17d059.json) · [upstream source ↗](https://huggingface.co/api/datasets/lmarena-ai/leaderboard-dataset)
- [licence · 88a46d472f35](/data/rankings/evidence/88a46d472f35274fb6b7a852c3d16c8ee8007d51b37f290161537bec26895ee4.txt) · [upstream source ↗](https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/raw/8b6deb69ace01ee8dcda9e46eba7a5e8a808ec53/README.md)
- [rows · 7c16d94881da](/data/rankings/evidence/7c16d94881dac8fe596fbd55013f2b4c5a5a88fef6d91768dbd234475fe1327a.parquet) · [upstream source ↗](https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/resolve/8b6deb69ace01ee8dcda9e46eba7a5e8a808ec53/agent/latest-00000-of-00001.parquet)

[Explore other model types →](/rankings/)[Read the benchmark guide →](/benchmarks/)

## Continue your investigation

 - [Understand the measurements](/benchmarks/)
- [Compare task performance](/best/)
- [Compare embedding models](/embeddings/)
- [Inspect ranking rules](/methodology/)