DI-engine provides an OpenDILab framework for reinforcement learning projects.
Open source
24 alternativesViZDoom
ViZDoom provides reinforcement learning environments where AI bots play Doom using screen information.
About ViZDoom
ViZDoom is a set of reinforcement learning environments in which AI agents play the 1993 game Doom using only visual information, the screen buffer. It is built on the ZDoom engine for game mechanics and is primarily intended for research in machine visual learning, particularly deep reinforcement learning.
Beyond screen images, the environment can provide depth buffer data, audio buffer access, automatic labeling of visible objects, and access to actors, map geometry, and in-game messages. Installation depends on platform and Python version. Audio features require OpenAL, and pre-built macOS wheels for version 1.3.0 and later are limited to Apple Silicon running macOS 14.0 or later.
Key features
- Python API with Gymnasium and Gym wrappers, plus a C++ API
- Depth buffer access for 3D vision
- Automatic labeling and categorization of game objects visible in the frame
- Synchronous and asynchronous single-player and multiplayer modes
- Custom scenario creation with visual editors and a scripting language
- Access to audio buffer, actor lists, map geometry, and in-game text messages
Best for: Researchers and students in visual machine learning and deep reinforcement learning who want Doom-based training environments.
Good for
- Supports Linux, macOS, and Windows.
- Lightweight, with a package size of a few MBs, and fast in sync mode, reaching up to 7000 steps per second single-threaded on a modern CPU.
- Offers both Python and C++ APIs, including Gymnasium and Gym wrappers.
Watch out for
- Audio features require OpenAL to be installed on the system.
- Pre-built macOS wheels for version 1.3.0 and later are available only for Apple Silicon running macOS 14.0 or later; Intel Mac users need version 1.2.4.
- Pre-built Linux wheels require Python 3.10 or later; otherwise pip builds from source, which needs a C++11 compiler, CMake 3.12+, Boost 1.54+, and SDL2.
ViZDoom at a glance
- Website
- github.com/mwydmuch/ViZDoom
- Type
- Open-source project
- Pricing
- Not confirmed. Check the site
- Platforms
- Linux, macOS, Windows
- Open source
- Not stated
- Language
- English
- Last checked
- Oct 6, 2026 · Online when checked
- Listed
- Oct 5, 2026
ViZDoom alternatives
Tools and sites you could use instead, ranked by how closely they match.
TF-Agents provides a TensorFlow library for contextual bandits and reinforcement learning.
Open source
24 alternatives
The somoclu project implements self-organizing maps in parallel across CPUs, GPUs, and clusters.
Open source
24 alternatives
OpenMetricLearning provides metric learning pipelines, models, and a model zoo.
Open source
24 alternatives
Use Learnergy, a Python library for training and working with energy-based machine learning models.
Open source
24 alternatives
Karate Club offers algorithms for unsupervised learning on graph data in Python.
Open source
24 alternativesPopular in ML Training & Optimization
Well-known listings from the same topic.
Edwardedwardlib.orgAerosolveaerosolve.airbnb.techDataturedatature.ioTogether AItogether.aiOptunaoptuna.orgCatBoostcatboost.ai
ViZDoom FAQ
How can I use ViZDoom on an Intel Mac?
Install version 1.2.4 with pip (pip install vizdoom==1.2.4), since pre-built wheels for version 1.3.0 and later are available only for Apple Silicon running macOS 14.0 or later.
Do I need extra software to use audio features?
Yes. Audio features require OpenAL. On apt-based Linux distributions it can be installed with libopenal-dev, and on dnf or yum-based distributions with openal-soft-devel.
What happens if no pre-built wheel exists for my Python version or platform?
pip will try to build ViZDoom from source. This requires a C++11 compiler, CMake 3.12+, Boost 1.54+, SDL2, and optionally OpenAL.
Is something wrong or out of date?
Tell us and an editor will check it.
Own this website?
Claim this listing
Claim the listing and suggest updates yourself. An editor checks every claim first.