Benchmark Almanac

PlayWorld

Genie 3 2.12 tops its leaderboard
Not in model cards

An agent player pursues objectives inside video world models. Gemini 3.1 Pro applies weighted rubrics for geometry, interaction, and state evolution across the resulting videos.

The judge is separate from the agent player. The repository's automatic-metric directory is empty, while the rubric evaluator is implemented.

Released
August 2026
Kind
Interactive world models
Level
Everyday
Scored
Graded by a model
Answers
Agent task
Tasks
All public
Measures
Capability
Language
English
In and out
Text, Video → Video
Interaction
Agentic
Environment
A simulator
Items from
Generated
Refresh
One fixed set
Run by
Anyone

In the authors’ words

Video world models simulate future states conditioned on current observations and user actions. Recent systems have demonstrated impressive video consistency and action controllability over long sequences. However, fairly comparing these interactive models remains challenging. In practice, a human player typically evaluates a world model by pursuing long-horizon objectives through interaction. For example, a user may turn around 360 degrees to see whether the environment remains consistent, or walk into the water and inspect whether realistic water ripples are generated. The action sequence required to achieve the same objective may vary substantially between models, making fixed action-conditioned evaluation unsuitable for cross-model comparison. To address this, we employ multi-modal Agent Players to interact with world models toward specified long-horizon objectives. Building on this paradigm, we introduce PlayWorld, a benchmark providing 171 scenarios, each with a specified objective. To evaluate performance thoroughly, we assess models along four core dimensions: geometry consistency, interaction fidelity, out-of-sight evolution, and insight evolution. In addition, we incorporate basic ability metrics for video quality and controllability. Experiments across nine state-of-the-art world models reveal that current models remain unreliable on long-horizon interactive objectives, particularly in maintaining spatial consistency and persistent state evolution. Code and data are available at https://github.com/kxding/PlayWorld.

Abstract of PlayWorld: Benchmarking World Models with Agent Players over Long-Horizon Objectives (arXiv:2608.13552, CC0)
9
Entries on its own leaderboard, 0 models we track

Genie 3 tops the leaderboard with 2.12

Best score per entry from the benchmark's own leaderboard, read on November 18th, 2026. Where a model ran with several settings or harnesses, we keep its best. Names in gray aren't in our model list: fine-tunes, agent systems, products and older models. They don't count toward leaders.

  1. 1 Genie 3 2.12
  2. 2 HappyOyster 1.92
  3. 3 LingBot-World2 1.82
  4. 4 LingBot-World 1.78
  5. 5 HY-World2 1.61
  6. 6 SANA-WM 1.48
  7. 7 Hunyuan-GameCraft 1.42
  8. 8 HY-WorldPlay 1.21
  9. 9 Matrix-Game-3.0 1.14

Source: Official PlayWorld leaderboard

#EntryScoreSettingDateSource
1 Genie 3 2.12 Official leaderboard
2 HappyOyster 1.92 Official leaderboard
3 LingBot-World2 1.82 Official leaderboard
4 LingBot-World 1.78 Official leaderboard
5 HY-World2 1.61 Official leaderboard
6 SANA-WM 1.48 Official leaderboard
7 Hunyuan-GameCraft 1.42 Official leaderboard
8 HY-WorldPlay 1.21 Official leaderboard
9 Matrix-Game-3.0 1.14 Official leaderboard

No lab model card reports it yet

We read the headline results of 540 launch posts from 57 labs, and none of them list this benchmark.

More in World models

All 53 in World models
  • AgentWorldBench In use Language world models predict the next environment observation after an agent action, in 7 domains from MCP and search to Android and desktop OS. A judge model scores each prediction against the real observation on five dimensions. GPT-5.4 58.3
  • IntPhys 2 In use Models watch simulated videos and judge whether each event is physically possible, testing object permanence, immutability, continuity and solidity. Scored by accuracy. Muse Spark Open 68.3%
  • QwenWorldBench Fading Models act as world models that simulate agent environments in 7 domains, such as terminal, web, and Android. A rubric judge scores five dimensions. Qwen3.7-Plus 62.1%
  • VPCT Fading Models view an image of a ball, ramps, and three buckets and predict which bucket the ball falls into. Scored by accuracy on 100 problems. Gemini 3 Pro 91%
  • ContPhy Fading Models infer physical properties such as mass and density of fluids and soft bodies from videos and predict their dynamics. Seed2.0 Pro 67.4%
  • RBench Only in system cards 650 image+text prompts for robot video generation (tasks and embodiments). MLLM judge scores task completion, physical plausibility, stability. Wan 2.6 60.7%

Data audit

All 8 checks pass. Every benchmark

Catalog record 2 pass
  • Pass Task access All tasks are public.
  • Pass Confidence Checked against a primary source.
Scores 2 pass 3 notes
  • Pass Independent runs Scores from the official leaderboard (9 entries).
  • Pass Models scored 9 entries on its leaderboard, none a model we track.
  • Note Newest score No scores yet.
  • Note Frontier models No models scored yet.
  • Note Headroom Scored as a plain number, with no fixed ceiling.
Leaderboard import 3 pass
  • Pass Importer We read the board on github.com: 9 entries at the last read, none a model we track.
  • Pass Last read November 18th, 2026.
  • Pass Read worked The last read worked.
Consistency 1 pass 1 note
  • Note Lab claims No model has both a lab number and an independent run.
  • Pass Known issues No published audit found problems.

How we check