Apple’s Foundation Models framework is a native Swift API for adding language-model capabilities to apps, with direct access to Apple’s on-device model and models running in Private Cloud Compute. It is designed for tasks ranging from summaries to agent-like workflows, and includes guided generation, tool calling, image understanding, and integration with Vision features such as OCR and barcode reading. The framework also supports server-hosted models, including Claude and Gemini, through a shared API; providers can offer models by implementing its Language Model protocol. Apple emphasizes privacy and offline operation for on-device use. Eligible developers in the App Store Small Business Program - with fewer than two million first-time downloads - can access Apple models on Private Cloud Compute without cloud API charges, though usage limits may apply.
The framework’s newer Dynamic Profiles let apps switch models, tools, instructions, temperature, and reasoning settings during a continuous session while retaining a shared transcript. This supports adaptable agents and skills. An Evaluations framework helps test prompt and model behavior under changing conditions, while the macOS 27 “fm” command-line tool enables scripting and model-workflow prototyping. Apple also offers an open-source utilities package with experimental skills and context-management helpers, plus a Python SDK for working with Python tools and evaluation packages. Developers can consult usage restrictions and documentation for integrating Apple’s cloud model. Server-side features may have daily limits that vary with demand and request complexity; expanded access is expected to become available for a fee.
Summary generated by AI from the linked article. hn.today is not affiliated with Hacker News or Y Combinator.
A Hacker News poster asks which practical AI skills individuals can learn online to benefit economically from LLMs and agents. They want specific, hands-on recommendations suited to someone with a moderately technical background, without requiring a degree. (news.ycombinator.com)
Former OpenAI researcher Will Depue claims hidden initialization patterns in six Chinese open-weight AI models match Nvidia Blackwell hardware, including systems restricted from export to China. He says the matches were reproduced on real chips and independently checked by researchers, though the claims are based on his forensic analysis. (twitter.com)
Antithesis built a system that beats NetHack on nearly 80% of game seeds within five minutes, across character builds. Successful runs take a median of under two minutes, advancing the company’s goal of creating smarter fuzzers and autonomous systems. (antithesis.com)
OpenCyclone is a free, open-source Android agent that uses your existing apps to carry out tasks through a hidden screen. It asks for approval before actions like sending messages and lets users connect their own AI models. (opencyclone.vercel.app)
Sam Bankman-Fried’s $500 million investment gave Alameda nearly 14% of Anthropic, a stake that would be worth about $37 billion at the AI company’s expected $1.5 trillion IPO valuation. The stake was sold in bankruptcy to repay FTX customers and creditors after Bankman-Fried’s businesses collapsed. (ft.com)
llama.cpp and ggml have reached v1.0, establishing stable public C APIs and clear versioning rules for breaking changes. The release strengthens the foundation for running local, multimodal AI across a wide range of hardware. (huggingface.co)
Today's best Hacker News stories, summarized and screenshotted, one email a day.