Swapping one import statement should not cost a developer half a day, but for Apple developers juggling local models, cloud APIs, and Apple’s own Foundation Models, it regularly does.

Three separate APIs for one job is two too many

Apple developers building AI-powered apps have had to maintain separate integration patterns for Core ML, MLX, OpenAI, Anthropic, and Apple Foundation Models simultaneously. Every new model experiment means new boilerplate, new error handling, and a fresh round of debugging before a single prompt gets answered.

One import replaces the entire switching cost

Introducing AnyLanguageModel is a Swift package that mirrors Apple’s Foundation Models API exactly, so developers replace one import line and immediately gain access to local MLX models, Core ML, Apple’s system model, and remote cloud providers through a single consistent interface. You write the same session and respond call regardless of whether the model runs on-device or in the cloud. The input is a prompt, the output is a response, and the provider is just a constructor argument.

The developers already losing time to this are obvious

• iOS engineers prototyping AI features who need to compare a local Qwen model against GPT-4o without rewriting session logic each time
• Indie Mac app developers who want privacy-first local inference as default with a cloud fallback when the task demands it
• Mobile AI teams at startups who need to ship fast and cannot afford a bespoke abstraction layer for every model provider they evaluate

Apple announced Foundation Models at WWDC 2025 for macOS 26 and iOS 26, and the developer window to ship polished AI features before the release cycle closes is narrow. Every week spent on integration plumbing is a week not spent on the actual product, and the teams that solve this first will have a meaningful head start on the App Store.

What you can do with it today

• Run a quantized MLX model locally with zero cloud dependency
• Switch providers by changing a single constructor argument in your code
• Test privacy-sensitive prompts on-device before deciding if cloud is needed
• Use Apple’s system model as a free fallback when API costs matter

Pricing not listed — check our directory.

The API surface mirrors Foundation Models closely, which means any future breaking changes Apple makes upstream could require AnyLanguageModel to catch up before your app compiles cleanly again.

LangChain covers multi-provider orchestration for Python and JavaScript stacks but has no native Swift story. Apple’s own Foundation Models framework supports only the system model, making AnyLanguageModel the only Swift-native option that bridges local and remote providers under one API today.

Apple’s AI developer story is being rewritten by third-party packages

The tooling gap between Apple’s official AI APIs and what developers actually need is closing fast, and the packages filling it now will become default dependencies before most teams notice the shift. We cover tools like this every Friday — subscribe here and we’ll send the best ones straight to you.