---
title: "I2A | Imagination-Augmented Agents | world-models.io"
canonical: "https://world-models.io/en/models/imagination-augmented-agents/"
description: "Imagination-Augmented Agents (I2A) use learned world models to imagine future trajectories for better decision-making in RL."
author: "Tyler D."
generated: "2026-10-30"
---

New: the Timeline is live. Track world model releases, papers, and benchmark updates in real time.

The Knowledge Hub for AI World Models

# Imagination-Augmented Agents (I2A)

I2A agents use learned environment models to imagine possible futures, extracting useful information from imagined trajectories to improve policy decisions.

robotics

model-based-rl

simulation

embodied-ai

## What is a world model?

A world model is an AI system that learns a predictive representation of how an environment changes over time. It can use observations, actions, or language to estimate future states and possible outcomes before an agent acts. World models support planning in model-based reinforcement learning, robotics, embodied AI, autonomous driving, and interactive simulation. This site catalogs these systems with model records, research papers, benchmark context, and source links. Start with the [world models database](https://world-models.io/en/world-models-database/), compare systems on the [leaderboard](https://world-models.io/en/leaderboard/), or read the [Performance Index methodology](https://world-models.io/en/methodology/performance-index/).

## Key Attributes

Static HTML snapshot of the model record for crawlers and no-JS readers.

| Attribute | Value |
| --- | --- |
| Model | Imagination-Augmented Agents (I2A) |
| Lab / Organization | DeepMind |
| Category | Model-Based RL |
| Subtype | Imagination-Augmented Policy |
| World Model Type | Imagination-augmented model-free agent |
| Primary Domain | Atari / Planning |
| Architecture | Model-free policy augmented with rollout encoder over imagined trajectories |
| Modality | Visual |
| Training Method | Joint training of environment model and imagination-augmented policy network |
| Status | foundational |
| Year | 2017 |
| Performance Index | 45/100 (medium confidence, v1.1) |

## About Imagination-Augmented Agents (I2A)

Main editorial body preserved directly in static HTML.

Imagination-Augmented Agents (I2A) combine model-free reinforcement learning with learned world models. The agent uses an environment model to generate imagined rollouts of possible futures, then aggregates these imagined trajectories as additional context for its policy. This allows the agent to benefit from planning-like behavior without fully committing to model-based control. I2A demonstrated that even imperfect learned models could improve policy quality when used as an augmentation rather than the sole basis for decision-making.

Imagination-Augmented Agents (I2A) is an imagination-augmented model-free agent developed by Google DeepMind in 2017 for atari / planning.

## Editorial Snapshot

Short extractable facts for answer engines and no-JS readers.

| Signal | Value |
| --- | --- |
| Definition | Imagination-Augmented Agents (I2A) is an imagination-augmented model-free agent developed by Google DeepMind in 2017 for atari / planning. |
| Short Description | An agent architecture that augments model-free policies with learned imagination rollouts from an environment model. |
| Benchmark Rows | 2 |
| FAQ Entries | 2 |
| Related Models | 4 |
| Related Guides | 0 |
| Related Research Topics | 1 |
| Last Updated | 2026-01-20 |

## Notable Features

Key capabilities associated with this model.

- • Combines model-free and model-based strengths
- • Tolerates imperfect environment models
- • Rollout encoder aggregates imagined futures
- • Improved Sokoban puzzle solving

## Use Cases

Representative applications attached to this model record.

Puzzle-solving (Sokoban)

Atari games

Planning-augmented RL

Hybrid model-based/model-free RL

## Strengths and Limitations

Balanced assessment surfaced in static HTML.

### Strengths

- • Robust to model errors
- • Improves model-free baselines
- • Conceptually elegant hybrid approach
- • Early demonstration of imagination value

### Limitations

- • Environment model quality still matters
- • Computational overhead of imagination rollouts
- • Superseded by Dreamer family

## Benchmarks

Published benchmark evidence attached to this model record.

| Benchmark | Metric | Result | Source |
| --- | --- | --- | --- |
| Sokoban | Solve Rate 87 % | Significant improvement over model-free | [Source](https://arxiv.org/abs/1707.06203) |
| Atari | Mean HNS 1.15 x baseline | Improved over baselines | [Source](https://arxiv.org/abs/1707.06203) |

## References and Citations

Primary references preserved in static HTML for citation extraction.

| Reference | Link |
| --- | --- |
| Weber et al., 2017. Imagination-Augmented Agents for Deep Reinforcement Learning. NeurIPS 2017. | [Open source](https://arxiv.org/abs/1707.06203) |

## Related Models

Nearby models linked from the current editorial record.

| Model | Category | World Model Type | Index v1.1 |
| --- | --- | --- | --- |
| [MuZero](https://world-models.io/en/models/muzero/) | Model-Based RL | Abstract learned dynamics + MCTS | 78/100 |
| [DreamerV3](https://world-models.io/en/models/dreamer-v3/) | Model-Based RL | Imagination-based dynamics model | 88/100 |
| [Predictron](https://world-models.io/en/models/predictron/) | Model-Based RL | Abstract internal dynamics model | 43/100 |
| [World Models (Ha & Schmidhuber)](https://world-models.io/en/models/ha-world-model/) | Model-Based RL | Generative latent dynamics model | 48/100 |

## Direct Comparisons

Side-by-side comparisons already connected to this model.

| Comparison | Matchup | Summary |
| --- | --- | --- |
| [Model-Based RL vs Model-Free RL](https://world-models.io/en/compare/model-based-rl-vs-model-free-rl/) | Model-Based RL vs Model-Free RL | Model-based RL learns a world model for imagination-based planning. Model-free RL learns directly from interaction without an internal model. Each approach has distinct strengths depending on the application domain. |

## Research Topics Referencing This Model

Connected research areas surfaced directly in static HTML.

| Topic | Summary |
| --- | --- |
| [Model-Based Reinforcement Learning](https://world-models.io/en/research/model-based-rl/) | What model-based reinforcement learning is, how world models enable imagination-based planning, and why Dreamer, MuZero, PlaNet, and TD-MPC2 matter. |

## Frequently Asked Questions

FAQ answers rendered directly into static HTML for extractable responses.

### How do I2As use imagination?

I2As generate imagined future trajectories using a learned environment model, encode them with a rollout encoder, and feed the encoded imaginations as additional inputs to the policy, augmenting rather than replacing model-free decision-making.

### Why not just use the model directly for planning?

Learned environment models are imperfect. I2As use imagination as supplementary information rather than relying on it entirely, making the system robust to model errors.

## Quick Answer

Short extractable summary preserved directly in static HTML.

- Imagination-Augmented Agents (I2A) is an imagination-augmented model-free agent developed by Google DeepMind in 2017 for atari / planning.
- Use this page when you need a fast read on how Imagination-Augmented Agents (I2A) fits into the model-based rl landscape, then validate the details in the benchmarks, citations, and related pages.
- A key strength surfaced in the editorial record is robust to model errors.

## Editorial Trust Signals

Editorial provenance and refresh policy preserved directly in static HTML.

**Published by** world-models.io editorial board.

**Lead editor** Tyler D. - Technical editor, methodology and benchmark analysis.

This model page synthesizes primary papers, official model pages, benchmark evidence, and related world-models.io context into a reference resource.

Each editorial page is assembled from primary sources, normalized into extractable summaries, checked for factual drift, and reviewed before publication or major refreshes. Last reviewed: 2026-01-20.

Pages are refreshed when a new paper, benchmark, release, architecture update, or stronger primary source materially changes the answer a reader or AI system should retrieve.

Each page links back to relevant primary sources and keeps a stable canonical URL so readers can verify claims, trace context, and reference the most up-to-date version. [See the editorial policy.](https://world-models.io/en/editorial-policy/)

Primary sources only

Last reviewed date visible

Methodology documented

Source links included

Correction channel published

Review date visible

## External Sources

Primary model and lab sources embedded in static HTML.

- [Imagination-Augmented Agents (I2A) paper](https://arxiv.org/abs/1707.06203) Google DeepMind primary research source
- [Google DeepMind official website](https://deepmind.google) Related lab

## References

1. [[1] Weber et al., 2017. Imagination-Augmented Agents for Deep Reinforcement Learning. NeurIPS 2017.](https://arxiv.org/abs/1707.06203)