Code | Datasets

Open Resources.

OpenEnvision organizes research artifacts as clean, reusable releases. Code, checkpoints, datasets, evaluation files, and documentation are treated as part of the same scientific record.

Resource Types

A release surface for every part of the research stack.

This page defines the resource categories OpenEnvision uses for public releases and highlights available open-source artifacts.

Code

Training, evaluation, data processing, and visualization code prepared with clear dependencies, runnable entry points, and minimal setup friction.

Models

Checkpoints, adapters, model cards, intended-use notes, and evaluation summaries for released systems.

Datasets

Curated multimodal, temporal, spatial, and embodied data with task definitions, metadata, splits, and responsible-use notes.

Benchmarks & Results

Evaluation protocols, leaderboards, result files, metric definitions, and comparison artifacts for reproducible analysis.

Featured Open Resource

WorldFoundry

Explore the public source repository and documentation for WorldFoundry. The code, setup instructions, technical references, and usage guides are available through the two official entry points below.

Open Source GitHub Documentation

View on GitHub Read the Docs

Dataset & Benchmark

WorldAtlas

Access the WorldAtlas dataset, benchmark, source code, and documentation through the four resource links below.

Dataset Benchmark Code Documentation

Dataset Benchmark Code Docs

Standards

What every public resource should make clear.

Version

Stable identifiers, update notes, and compatibility expectations.

Scope

Intended use, task definitions, data coverage, and known limitations.

Evidence

Evaluation settings, metrics, result files, and comparison context.

Reuse

License, citation, file layout, and practical loading instructions.