Search papers, labs, and topics across Lattice.
5
32
9
5
Frontier MLLMs exhibit strong local perception but falter on long-horizon audio-visual reasoning, with a single benchmark exposing specific failure modes like identity drift and cross-modal misalignment.
Finally, a video generation model that lets you puppeteer the camera independently of the scene, unlocking creative control previously impossible.
Language models are increasingly doing their real work in the "invisible" latent space, not the tokens we see.
Skip the costly VFX pipeline: EffectMaker lets you create customized visual effects with a reference video and a multimodal LLM, no per-effect fine-tuning needed.
The largest open-source image generative model to date, HunyuanImage 3.0, achieves state-of-the-art performance using a Mixture-of-Experts architecture and native Chain-of-Thoughts schema.
