PulseAugur
EN
LIVE 18:57:33
中文(ZH) 智源大会 | 天工AI重新定义世界模型,公布Matrix-Game 3.5 最新技术突破

Kunlun Wanwei's Matrix-Game 3.5 advances world models with joint state-action training

At the 8th Zhiyuan Conference, Kunlun Wanwei's Skywork team unveiled Matrix-Game 3.5, a significant advancement in world models. The new framework proposes joint training for state and action generation, moving beyond simple next-frame prediction. This iteration also introduces enhanced long-term memory capabilities and expands its application from game environments to real-world scenarios, with a planned release in July 2026. AI

IMPACT This advancement in world models could accelerate progress in robotics and embodied AI by enabling more realistic and interactive simulations.

RANK_REASON The cluster describes a new model release (Matrix-Game 3.5) with technical breakthroughs and a planned release date, presented at a major AI conference. [lever_c_demoted from frontier_release: ic=1 ai=1.0]

Read on 量子位 (QbitAI) →

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →

Kunlun Wanwei's Matrix-Game 3.5 advances world models with joint state-action training

How we ranked this

Signal score
0 / 100
Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket
Significant
The cluster describes a new model release (Matrix-Game 3.5) with technical breakthroughs and a planned release date, presented at a major AI conference. [lever_c_demoted from frontier_release: ic=1…
Source corroboration
Single-source cluster
Only one publisher covered this so far. Single-source stories can still rank when the publisher is high-authority, but they lack cross-source corroboration.
Topics
model release, paper
Editorial topic classification. Feeds into how the story surfaces on /topic/<slug> hub pages and into the per-entity coverage mix.
AI-industry relevance
High
Clearly on-topic for AI-industry coverage.
Story freshness
77 days old
Aged out of breaking-news scoring windows; ranking reflects the durable signal from the full source set.

Full methodology in our editorial standards.

COVERAGE [1]

  1. 量子位 (QbitAI) TIER_1 中文(ZH) · 量子位的朋友们 ·

    Zhiyuan Conference | Tiangong AI Redefines World Models, Announces Latest Technological Breakthroughs in Matrix-Game 3.5

    昆仑万维将持续深耕,步履不停