Running 41 Don't Train the Model, Evolve the Harness πΏ 41 Evolving an agent's harness, not its model, on Harvey's LAB
Running on Zero Agents 165 Gemma 3 12b It π₯ 165 Chat with an AI that understands text, images, and videos
Running on Zero Agents Featured 95 Gemma 4 12B IT β‘ 95 Chat with an AI using text, images, audio, or video
Paused Agents Featured 57 PROMETHEUS v1.0 β World Model Interactive Demo π₯ 57 World-first embodied AI world model
Running on Zero MCP 2k Wan2.2 14B Fast Preview π 2k generate a video from an image with a text prompt
VibeThinker-3B: Exploring the Frontier of Verifiable Reasoning in Small Language Models Paper β’ 2606.16140 β’ Published 23 days ago β’ 122
Running on Zero Agents 5 VibeThinker 3B π» 5 Generate detailed answers with stepβbyβstep reasoning
LeWorldModel: Stable End-to-End Joint-Embedding Predictive Architecture from Pixels Paper β’ 2603.19312 β’ Published Mar 13 β’ 50
Causal-JEPA: Learning World Models through Object-Level Latent Interventions Paper β’ 2602.11389 β’ Published Feb 11 β’ 11
ThinkJEPA: Empowering Latent World Models with Large Vision-Language Reasoning Model Paper β’ 2603.22281 β’ Published Mar 23 β’ 20
Running on Zero Agents 17 World Tracing Demo π 17 Multilayer-geometry 3D from a single image or 16-frame clip