OmniBehavior - Video Binary Behaviors: leaderboard

OmniBehavior (2026) tests LLMs as simulators of real users: given the profile and long, real behaviour history of a user across the Kuaishou video, live-streaming, ads and e-commerce scenarios, the model predicts what the user does next. F1 on binary engagement decisions such as liking or following in the short-video scenario, in percent; higher is better.

Metric: F1 (%) on binary engagement decisions (such as like, follow, share) in the short-video scenario, for predicting the real behaviour of 200 Kuaishou users (three months of interleaved video, live-stream, ads and e-commerce traces, 8,143 actions per user on average) from the user profile, history and current scenario, 6,000 balanced tasks, 32k context window, temperature 0.1; higher is better. Source: arxiv.org. Saturation forecast: Around October 2028. 11 models tracked. Updated 7 Oct 2026.

Top models

#ModelScoreOverall rank
1Claude Opus 4.533.05#160
2GPT-5.231.54#185
3GPT-4o27.88#751
4GLM-4.726.86#380
5Claude Sonnet 425.29#410
6Kimi K2 090523.3#570
7Claude Haiku 4.522.84#583
8Gemini 3 Flash22.09#175
9OmniBehavior DeepSeek-V3 (checkpoint unspecified)21.42
10Claude Sonnet 4.518.85#261
11Qwen 3 235B A22B 2507 Instruct18.26#616

No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=omnibehavior-video-binary-behaviors · How It Works · Data refreshed daily, snapshot 2026-11-21.