BotBeat
...
← Back

> ▌

AnthropicAnthropic
UPDATEAnthropic2026-06-08

Anthropic Brings Claude to Apple's Foundation Models Framework

Key Takeaways

  • ▸Claude is now accessible through Apple's Foundation Models framework via a new Swift package
  • ▸Developers can create hybrid workflows combining on-device models for simple tasks with Claude for complex reasoning and analysis
  • ▸The integration supports streaming, tool calls, web search, code execution, and structured responses
Source:
Hacker Newshttps://claude.com/blog/claude-for-foundation-models↗

Summary

Anthropic has released a new Swift package enabling Apple developers to integrate Claude with Apple's Foundation Models framework. The integration allows developers to create hybrid AI workflows where on-device models handle simple tasks like summarization and extraction, while Claude handles more complex reasoning, multi-step workflows, code generation, and web search capabilities.

The Foundation Models framework enables developers to tap into models natively from Swift and return typed values through guided generation. By combining Apple's on-device models with Claude's advanced capabilities, developers can now create more sophisticated applications. The integration is available on iOS 27, iPadOS 27, macOS 27, visionOS 27, and watchOS 27.

Examples of use cases include journaling apps that generate daily prompts locally before asking Claude to find patterns across months of entries, and study apps that define terms locally then ask Claude more complex follow-up questions. The integration handles streaming, tool calls, and structured responses directly into SwiftUI views, making it seamless for developers to implement hybrid AI experiences.

  • Available on all major Apple platforms: iOS 27, iPadOS 27, macOS 27, visionOS 27, and watchOS 27
Large Language Models (LLMs)Generative AIPartnerships

More from Anthropic

AnthropicAnthropic
PRODUCT LAUNCH

DingDuff: Claude-Powered Legal Research Tool Launches with Tip-Jar Model

2026-07-23
AnthropicAnthropic
RESEARCH

Anthropic Advances LLM Guardrails with Multilingual, Context-Aware Evaluation Framework

2026-07-23
AnthropicAnthropic
RESEARCH

Anthropic Interpretability Paper Reveals LLMs Have Measurable Internal 'Workspace'

2026-07-23

Comments

Suggested

OpenAIOpenAI
RESEARCH

OpenAI's Advanced Models Autonomously Breach Sandbox, Hack Hugging Face in Unprecedented AI Security Incident

2026-07-24
Australian GovernmentAustralian Government
POLICY & REGULATION

Australia Mandates Energy Independence for AI Datacenters, Requires Content Creator Consent

2026-07-24
OracleOracle
FUNDING & BUSINESS

Oracle's $300B OpenAI Bet Backfires: 21,000 Layoffs and Credit Downgrade Jeopardize Computing Deal

2026-07-23
← Back to news
© 2026 BotBeat
AboutPrivacy PolicyTerms of ServiceContact Us