Meet LingBot-World-Infinity: An Open Causal World Model With An Agentic Harness
Ant Group's new 14B world model runs 60 minutes uninterrupted. Here's why that matters for agents.
Why it matters
Ant Group's Robbyant unit released a causal video generation model designed to solve long-horizon drift in interactive world simulators—a critical bottleneck for embodied AI agents. The MoBA attention mechanism and Director-Pilot agentic architecture represent a meaningful capability jump, though limited release scope (no quantitative benchmarks, non-commercial license) constrains immediate industry impact.
The key facts
8 to knowLingBot-World-Infinity: 14B parameter causal video generation model
Mixture of Bidirectional and Autoregressive (MoBA) attention mask architecture
60-minute uninterrupted session across 20 scenarios demonstrated
Director-Pilot agentic harness: VLM proposes events, Diffusion Transformer renders
Targets long-horizon drift failure mode in world models
Released under non-commercial CC BY-NC-SA 4.0 license
One checkpoint, 480P reference script; no deployment code or quantitative benchmarks provided
Source: Robbyant (Ant Group embodied-intelligence unit)
Go to the source
MarkTechPostmarktechpost.com
Publisher excerpt: Robbyant, Ant Group's embodied-intelligence unit, has released LingBot-World-Infinity (LingBot-World 2.0). It is a 14B causal video generation model that behaves as an interactive world simulator. The core idea is the Mixture of Bidirectional and Autoregressive (MoBA) attention mask, paired with…