HaiAI123

Curated Global AI Tools Directory

SurgeIndustry update
↑ 220%

Kunlun Releases Matrix-Game 4.0, the World's First Consumer-Grade GPU Video LLM Game, and Open Sources It

10/15/2026 — 10/15, 16:22·3 sources (all)·3 reports (all)

Key takeaways

  • Kunlun has released Matrix-Game 4.0, the world's first consumer-grade GPU video LLM game, achieving real-time gameplay at 480p, 60 FPS, with…

Background & analysis

On October 15, 2026, Kunlun Wanwei released Matrix-Game 4.0, presenting it as the world's first game built on a large video model that can run on a single consumer-grade graphics card. Chinese tech media ITHome covered the launch the same day. According to the company, the game delivers a real-time experience at 480p and 60 FPS on one NVIDIA RTX 5090, with latency below 150 milliseconds between a key press and the on-screen response.

In its announcement, Kunlun Wanwei framed the release as a new paradigm for game world models. The system pairs a game engine, which maintains the world state and the interaction rules, with a video generation model that renders detailed visuals in real time. The company says Matrix-Game 4.0 is the first to run a complete loop in which code generates the game logic, structured imagery supplies structural constraints, and the video model handles real-time rendering — a setup it presents as a practical technical path for game prototyping, personalized customization, and UGC.

Matrix-Game 4.0 was open-sourced at launch: the code is available on GitHub at https://github.com/SkyworkAI/Matrix-Game.git, and a project website went live alongside it. Kunlun Wanwei also laid out what comes next. The planned updates include deploying a refiner to push image resolution higher, exploring multiplayer sessions in which every player sees a consistent perspective, providing finer-grained control over the generated visuals, and extending support to more game genres and scenes. All of these are framed as future work; as of the same-day coverage, the launch and the open-sourcing remain the latest developments.

AI-generated from 1 reports · updated 5 hours ago

Latest turnSkywork (Kunlun) announced Matrix-Game 4.0 on Oct 15, describing it as the first video-model-driven game that runs on a single consumer GPU, released alongside open weights. A game engine maintains world state and interaction rules while a video generation model renders the visuals in real time, delivering 480p at 60 FPS on an NVIDIA RTX 5090 with under 150ms input-to-display latency.

24-hour heatHeat index 262 · peak 282 · 2h ago
24 hours agonow

Reports on this story headlines open the original

Today
  1. Skywork (Kunlun) announced Matrix-Game 4.0 on Oct 15, describing it as the first video-model-driven game that runs on a single consumer GPU, released alongside open weights. A game engine maintains world state and interaction rules while a video generation model renders the visuals in real time, delivering 480p at 60 FPS on an NVIDIA RTX 5090 with under 150ms input-to-display latency.

    36氪AI score 60
  2. Kunlun Tech launched Matrix-Game 4.0 on Oct 15, 2026, calling it the first video-model game that runs on a single consumer GPU. A game engine maintains world state and interaction rules while the video model renders 480p at 60 FPS in real time, with under 150 ms input latency on an NVIDIA RTX 5090. The project is open-sourced on Hugging Face under Skywork/Matrix-Game-4.0.

    雷峰网AI score 63
  3. Kunlun Group released Matrix-Game 4.0, the world's first video model game running on a single consumer GPU, achieving real-time 480p 60 FPS gameplay with under 150ms latency. The project is open-sourced and will improve resolution and multiplayer experience in future iterations.

    IT之家AI score 77

Other stories people are talking about

How is heat calculated?About the method

Heat counts how many independent sources covered a story in the last 48 hours: one source counts once no matter how many posts it published, decaying with a 24-hour half-life. What ranks first is what many people are talking about.

This page aggregates public feeds. Headlines and summaries are machine-organized and remain the property of the original authors; verify important facts at the source. If you believe a headline or summary infringes your rights, email the contact address in the footer with the page URL and basis of your claim, and we will remove or replace it after review.

Surge
Discussion rising fast
New
First report within 6 hours
Rising
Still gathering discussion

Back to AI industry updates →