# etched-ai/open-oasis

Inference script for Oasis 500M

Repository: https://github.com/etched-ai/open-oasis
Canonical: https://ross.abutalabs.com/products/open-oasis
Language: Python
License: MIT
License Family: permissive
Last push: 2024-11-08T23:28:40+00:00

## Health v2 (maintenance only)
Score: 22/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 0, release rhythm 35, longevity 47
- inputs: {"age_days": 671, "days_push": 663, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 2123, forks 186 (observed 2026-08-28T04:06:16.722511+00:00)

## What it is
Inference code and model weights for Oasis 500M, an interactive world model from Decart and Etched that generates gameplay video autoregressively conditioned on keyboard actions. It uses diffusion transformers (DiT) with a ViT VAE to produce action-conditional frame generation from a prompt frame.

## Use cases
- generate gameplay video from keyboard input
- run action-conditional video frame generation
- experiment with interactive world models
- generate video from a custom image prompt
- research autoregressive diffusion transformers

## When to choose
- you want to run or experiment with the Oasis 500M world model locally
- you need open weights for action-conditional video generation research
- you have a CUDA GPU and want to reproduce the demo gameplay generation

## When to avoid
- you need the full-quality Oasis model rather than the downscaled 500M version
- you lack a CUDA GPU or the memory to run diffusion transformer inference
- you want a polished game or end-user application rather than an inference script

## Facets
- artifact type: library
- maturity: experimental
- function: machine-learning, deep-learning, llm-inference, video-processing, simulation
- domain: artificial-intelligence, deep-learning, gaming-tools, simulation
- platform: python
- tags: world-model, diffusion-transformer, interactive-video-generation, gameplay-generation, model-weights, gpu, linux

## Member repositories
- etched-ai/open-oasis (main) score 22

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:06:16.722511+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T02:52:49.963465+00:00, confidence not recorded.
  - readme: https://github.com/etched-ai/open-oasis (fetched 2026-08-28T04:06:16.722511+00:00, sha f31543f21294)
- Data as of 2026-08-30T08:39:29.467469+00:00.