The 14th International Conference on Learning Representations (ICLR) wrapped up in Rio de Janeiro, drawing AI researchers from China, India, Europe and the U.S. Among the tech giants on the expo floor, Apple kept a prominent booth. There it demonstrated an open‑source model that converts 2D images into navigable 3D spaces within seconds, a tool the company called apple sharp 9to5mac.

Apple also highlighted its MLX framework, an open‑source stack that runs large‑language‑model inference on Apple Silicon, positioning the chip line as a viable alternative to traditional GPU farms. The booth doubled as a hiring hub, with iPads scanning QR codes for on‑the‑spot applications. Researchers presented dozens of Apple‑authored papers, including Federico Danieli’s ParaRNN and Kunal Talwar’s data‑pruning study.

With Amazon, Google, Microsoft and Meta also sponsoring ICLR, Apple’s showcase signals a renewed push to integrate its silicon‑centric AI stack into mainstream research pipelines. By open‑sourcing both the 3D conversion model and the MLX runtime, the company invites external validation while courting talent directly on the conference floor. Apple’s presence reaffirmed its ambition to be a first‑class AI platform provider.