At WWDC we introduce a new generation of LLMs developed to enhance the Apple Intelligence features. We also introduce the new Foundation Models framework, which gives app developers direct access to the on-device foundation language model. machinelearning.apple.com/research/apple…

🎉 Our paper UniVG is accepted to ICCV 2025! 📝 Paper: arxiv.org/pdf/2503.12652 #ICCV2025 #ImageEditing

In this report we describe the 2025 Apple Foundation Models ("AFM"). We also introduce the new Foundation Models framework, which gives app developers direct access to the on-device AFM model. machinelearning.apple.com/research/apple…

Post-training: Qibin Chen Ke Ye Dong Yin Violet Yao Zhaoyang Xu Multimodal: Bowen Zhang lyttonhao Yinfei Yang Zhe Gan Peter Grasch Framework/API: Zirui Wang Guoli Yin Richard Wei Erik Hornberger Project management: @Yuliyang19 sydevans

Vision tokenizers are stuck in 2020🤔while language models revolutionized AI🚀 Language: One tokenizer for everything Vision: Fragmented across modalities & tasks Introducing AToken: The first unified visual tokenizer for images, videos & 3D that does BOTH reconstruction AND

Apple presents Manzano: Simple & scalable unified multimodal LLM • Hybrid vision tokenizer (continuous ↔ discrete) cuts task conflict • SOTA on text-rich benchmarks, competitive in gen vs GPT-4o/Nano Banana • One model for both understanding & generation • Joint recipe:

Check out our team’s new work — Manzano 🍎🌲

Checked out Apple’s latest Flagship project! While focused on AToken, I admired the compute and resources on this flagship project, and the result is also amazing🎉

Just read the Manzano paper. They really wrote it in Pages, Sans font, plots in Numbers O.ô But after getting past that superficial weirdness, it's nice work. I like this figure, which clearly illustrates "big model smell" I also like it when Parti did that already, see below.

Checkout our recent Flagship research project Manzano, a simple, scalable unified multimodal model for image understanding and generation. 🚀🚀 Why 𝗠𝗮𝗻𝘇𝗮𝗻𝗼? - It is a Spanish word for an apple tree 🍎🌳 - Just one unified hybrid vision tokenizer with support for both

Apple presents Ferret-UI Lite Lessons from Building Small On-Device GUI Agents

🤔 Knowledge stored in multimodal LLM weights can be inherently limited. How can we empower multimodal LLMs with multimodal web search? 💡 In DeepMMSearch-R1, we aim to train a multimodal search agent capable of performing on-demand, multi-turn web searches and dynamically

Apple presents DeepMMSearch-R1 Empowering Multimodal LLMs in Multimodal Web Search

💡 Computer-use agents (CUAs) rely exclusively on primitive actions (click, type, scroll) that require lengthy execution chains, which can be cumbersome and error-prone. How to improve this? 🔥 🔥 In our native agent UltraCUA, we advocate the idea of "hybrid action" --

🎁🎁 We release Pico-Banana-400K, a large-scale, high-quality image editing dataset distilled from Nana-Banana across 35 editing types. 🔗 Data link: github.com/apple/pico-ban… 🔗Paper link: arxiv.org/abs/2510.19808 It includes 258K single-turn image editing data, 72K multi-turn

UltraCUA: A Foundation Model for Computer Use Agents with Hybrid Action UltraCUA is a foundation model that enhances computer-use agents by integrating low-level GUI actions with high-level programmatic tool calls through a hybrid action mechanism, significantly improving their