Phi-4-reasoning-vision
Open-weight 15B multimodal model for thinking and GUI agents
Open-weight 15B multimodal model for thinking and GUI agents
Redefining the Nothing aesthetic with a metal unibody
7B open model mixing transformers and linear RNNs
Codex now runs natively on Windows with secure sandbox
Frontier open-source MoE model built for OpenClaw agents
Best-in-class intelligence for your high-volume workloads
0.8B-9B native multimodal w/ more intelligence, less compute