Apple Foundation Models 2024

Name: Apple Foundation Models 2024
Author: Apple

AppleProprietaryPending Human Review

Apple’s first foundation language-model generation comprises a roughly 3B on-device model and a larger server model for Private Cloud Compute. Both use decoder-only Transformers with grouped-query attention and are adapted to everyday writing, summarization, extraction, and short dialogue through task-specific adapters. The models were publicly described at WWDC 2024; individual Apple Intelligence features rolled out later. Compression, efficient attention, and Apple-silicon optimization support local inference and cloud escalation. These proprietary language models are distinct from the image generators elsewhere in Apple Intelligence. Public model weights and a general hosted inference endpoint were not released.

2024-06-10

Approximately 3B on-device; server size undisclosed

Decoder-only Transformers with grouped-query attention

Proprietary

Compare with other models

Specifications

Parameters: Approximately 3B on-device; server size undisclosed
Architecture: Decoder-only Transformers with grouped-query attention
License: Proprietary
Type: text
Modalities: text

Benchmark Scores

Advanced Specifications

Model Family: Apple Foundation Models
API Access: Not Available
Chat Interface: Not Available
Variants: AFM-on-device (2024)AFM-server (2024)
Hardware Support: Apple silicon

Capabilities & Limitations

Capabilities: summarizationtext refinementinformation extractionshort dialogueon-device inference
Known Limitations: Published generation date does not imply simultaneous availability of all product featuresNo downloadable weights or general hosted model endpointFeature access depends on compatible Apple hardware and software
Notable Use Cases: writing assistancenotification summariesprivate text processing

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