Dispatches
News and developments shaping story generators, world models and the future of interactive media.
Meta releases Muse Ember, a small model built for agent subtasks
Meta launched Muse Ember on 9 November 2026: a low-cost model with a 1M-token context, priced at $0.10 per million input tokens, for high-volume agent subtasks.
12 Nov 2026 · Meta AI Research blog
WARC world model learns unseen ARC-AGI-3 game rules while it plays
WARC, a 35M-parameter world model, keeps training during play and learns most rules of unseen ARC-AGI-3 games. Code and weights are out under CC0.
10 Nov 2026 · arXiv preprint
World Labs' Frameboy streams its Atlas world model in real time
World Labs has released Frameboy, a research preview that streams scenes from its Atlas world model to the browser at up to 15fps as you steer the camera.
5 Nov 2026 · World Labs blog
Adobe and KAIST unveil MoviEdit, an AI tool for multi-shot video edits
Adobe Research and KAIST describe MoviEdit, a two-stage AI video editor that keeps edits consistent across scene cuts in clips up to 25 seconds long.
4 Nov 2026 · arXiv paper (2611.02737)
Tencent researchers unveil Wizard, an agentic world model for games
Wizard, from Tencent researchers, lets an agent edit a game world's explicit state and renders the result as real-time video at 24 FPS. Preprint out 29 October 2026.
2 Nov 2026 · arXiv paper (2610.40339)
Tencent open-sources Hy-Nomas, an 8B interactive world model
Tencent Hunyuan released Hy-Nomas, an Apache 2.0 video world model steered by text and camera moves across multiple turns, plus a 280-case benchmark.
30 Oct 2026 · Tencent-Hunyuan GitHub repository
Google DeepMind unveils Genie 4, a real-time world model with audio
Google DeepMind's Genie 4 research preview generates interactive worlds with native sound and dialogue that you can re-steer mid-session with text prompts.
28 Oct 2026 · Google DeepMind blog
Gradium releases Phonon, a family of tiny open-weight text-to-speech models
Gradium has released open weights for Phonon, 31m and 40m parameter TTS models for CPUs, phones and browsers. Only the English 31m model allows commercial use.
20 Oct 2026 · Gradium blog
Skywork open-sources Matrix-Game 4.0, a neural renderer for playable low-poly games
Matrix-Game 4.0 keeps gameplay in a real game engine and uses a streaming video model to repaint its low-poly frames. Code, games and weights are Apache 2.0.
15 Oct 2026 · Project GitHub
AMD agrees to acquire World Labs for about $8.2 billion
AMD has agreed to buy World Labs, the Fei-Fei Li-led maker of the Marble 3D world model, in an all-stock deal expected to close by the end of 2026.
28 Sept 2026 · AMD Newsroom
Unity releases an official plugin that teaches coding agents how to work in Unity
Unity's first-party plugin for Claude Code installs 29 skills written by its own feature teams, and the same skills now ship for OpenAI's Codex and xAI's Grok Build.
9 Sept 2026 · Unity blog
Runway shows GWM Worlds 2, a real-time world model driven by text actions
Runway's GWM Worlds 2 research preview generates interactive 720p video with audio at 24 fps, steered by text actions aimed at characters or the whole scene.
3 Sept 2026 · Runway Research
Unreal Engine 5.8 adds an experimental MCP plugin, and Epic puts LLMs into the plan for UE6
Epic released Unreal Engine 5.8 with an experimental MCP plugin that lets any language model inspect and act on a project, and said UE6 will build on the approach.
17 Jun 2026 · Unreal Engine blog