Apple Intelligence On-Device Models by Apple Machine Learning Research
Last refreshed 2026-04-15. Next refresh: weekly.
Details
About
Apple's on-device large language model family, a core component of Apple Intelligence, is designed for efficient and private operation directly on iPhones, iPads, and Macs. These models are optimized for speed and power efficiency, leveraging Apple Silicon to minimize memory footprint and power consumption 13. The family includes models of varying sizes, allowing for adaptability to different tasks and device capabilities 3. While specific details about the individual models within the family are not fully public, Apple's research indicates a focus on responsible AI development, including measures to mitigate bias and protect user privacy 1. The models power a range of features, including text summarization, rewriting, and proofreading, as well as enhanced Siri capabilities and intelligent image processing 1. Apple's commitment to on-device processing ensures that user data remains secure and private, minimizing reliance on cloud-based solutions 1. The models are continuously improved and expanded upon, with additional features and language support planned for future releases 1.
Decision facts
- Best fit
- General model comparison
- Capability starting point
- Apple On-Device
- Lowest tracked input
- Not tracked
- Closest related family
- OpenELM
Current Variants
Use-when guidance is based on each model's tracked capabilities, context window, release date, and replacement status.
| Model | Use when | Released | Signals | Status |
|---|---|---|---|---|
| Apple On-Device | Use when the workload needs 3B parameters. | 2024-06 | 3B parameters | Current |
Release Timeline
1 release groupSpecifications(1 models)
| Model | Released | Parameters |
|---|---|---|
| Apple On-Device | 2024-06 | 3B |
