New: the Timeline is live. Track world model releases, papers, and benchmark updates in real time.
world-models.io
The Knowledge Hub for AI World Models

Model-Based RL

World models evaluated through downstream control and sample efficiency on shared reinforcement-learning benchmarks.

robotics model-based-rl simulation embodied-ai

Comparable protocols

World models evaluated through downstream control and sample efficiency on shared reinforcement-learning benchmarks.

Atari 100K v1

Results require a documented task set, environment and ROM version, observation pipeline, seed count, interaction budget, Mean HNS, and human-normalized unit before they are comparable.

ModelMetricResultEvidence
dreamer-v3Mean HNS2.01 x humanAuthor reported
irisMean HNS1.046 x humanAuthor reported
diamondMean HNS1.56 x humanAuthor reported

DeepMind Control Suite v1

Results require a documented DMControl task set, environment version, observation settings, seed count, and mean-return metric before they are comparable.

ModelMetricResultEvidence
dreamer-v3Mean Return901 avg returnAuthor reported

Additional reported evidence

These source-linked author reports are retained for traceability but excluded from comparisons because they do not match a registered shared protocol.

ModelReported benchmarkResultWhy excluded
dreamer-v2Atari 200M1 x humanSource benchmark name does not match a registered public protocol.
planetDMControl Suite10 x more efficientSource benchmark name does not match a registered public protocol.
muzeroAtari7.31 x humanSource benchmark name does not match a registered public protocol.
td-mpc2DMControl (30 tasks)879 avg returnSource benchmark name does not match a registered public protocol.
imagination-augmented-agentsAtari1.15 x baselineSource benchmark name does not match a registered public protocol.
value-prediction-networkAtari1.08 x baselineSource benchmark name does not match a registered public protocol.