Apple Foundation Models 2025

Name: Apple Foundation Models 2025
Author: Apple

AppleProprietaryPending Human Review

The 2025 Apple Foundation Models generation adds image understanding, improved reasoning and tool use, and support for 15 languages. It pairs an approximately 3B on-device model using KV-cache sharing and 2-bit quantization-aware training with a server model using Parallel-Track Mixture of Experts and interleaved local/global attention. The Foundation Models framework exposes the on-device language model through Swift, with guided generation, tools, and task adapters; that interface should not be assumed to expose every multimodal capability described in the research report. The models remain proprietary, optimized for Apple silicon and Private Cloud Compute, with no public weight downloads.

2025-06-09

Approximately 3B on-device; server size undisclosed

On-device dense Transformer; server Parallel-Track Mixture-of-Experts Transformer

Proprietary

Compare with other models

Specifications

Parameters: Approximately 3B on-device; server size undisclosed
Architecture: On-device dense Transformer; server Parallel-Track Mixture-of-Experts Transformer
License: Proprietary
Type: multimodal
Modalities: textimage

Benchmark Scores

Advanced Specifications

Model Family: Apple Foundation Models
API Access: Available
Chat Interface: Not Available
Multilingual Support: Yes
Variants: AFM-on-device (2025)AFM-server (2025)
Hardware Support: Apple silicon

Capabilities & Limitations

Capabilities: multilingualimage understandingtool useguided generationon-device inference
Known Limitations: Local framework access depends on compatible Apple devices and softwareOn-device model is optimized for app tasks rather than general world-knowledge chatAdapters require retraining for new base-model versionsNo general hosted inference endpoint or public weights
Notable Use Cases: local app assistantsstructured extractionprivate summarization
Function Calling Support: Yes
Tool Use Support: Yes

Resources

Related Models

AFM 3 Cloud Pro

Apple

AFM 3 Cloud Pro is a third-generation Apple Foundation Model published at WWDC26 as part of Apple Intelligence. The highest-capability server tier is built with Google and runs on NVIDIA GPUs through an extension of Private Cloud Compute for demanding reasoning and agentic tool use. Access is integrated into Apple operating-system features rather than provided as freely downloadable weights. Availability depends on the device, software version, and supported feature. The June 8 date records the published generation; it does not imply that every consumer feature became generally available that day.

AFM 3 Cloud

Apple

AFM 3 Cloud is a third-generation Apple Foundation Model published at WWDC26 as part of Apple Intelligence. The server workhorse advances Apple’s PT-MoE architecture for multimodal reasoning while running on Apple silicon in Private Cloud Compute. Access is integrated into Apple operating-system features rather than provided as freely downloadable weights. Availability depends on the device, software version, and supported feature. The June 8 date records the published generation; it does not imply that every consumer feature became generally available that day.

AFM 3 Core Advanced

Apple

AFM 3 Core Advanced is a third-generation Apple Foundation Model published at WWDC26 as part of Apple Intelligence. Its prompt-level expert selection loads a subset of weights from flash into DRAM, enabling a larger local multimodal model within constrained memory. Access is integrated into Apple operating-system features rather than provided as freely downloadable weights. Availability depends on the device, software version, and supported feature. The June 8 date records the published generation; it does not imply that every consumer feature became generally available that day.

Typemultimodal

Parameters20B total, 1–4B active

2026-06-08

Proprietary

Details Compare