Apple Foundation Models

“The Foundation models keep a surprisingly decent set of data available for fast filtering by a local LLM.”

“well-balanced and relatively small local-inference models that ship with every mac”

“I've been impressed even with how Siri's been doing with the Apple Foundation Models in MacOS/iOS 27 given how small they are.”

“The Apple Foundation models SDKs are surprisingly cool.”

“Apple's local foundation models; they're downright eclipsed by the quality of Gemma”

“without having you download a second LLM and double your memory usage”

“fairly good plan for supplying a common API and default on device models”

“They were wrong when their on-device model was way behind. They still might be right in the long term.”

“my conversions to paying user shot up like 3x! But it really isn't a replacement obviously to claude”

“This seems smart. Apple, despite not really leading in AI themselves, are right on the hot path”

“I'm happy with Apple introducing this abstraction. my main concern was with local models.”

“It's only on Apple devices this API is used, so they split up the market by not letting developers use the same system if they want things to work on iOS”

“the Apple approach of swapping the experts only rarely is interesting, though it likely degrades the model a lot”

“it performs tool calling accurately and it is a very small model”

“It's not very good at world-knowledge recall, but can describe the submitted image well enough.”

“The idea of a system wide (and platform wide) on device model being a core part of OS APIs is very appealing.”

“The ability to tap into on-device foundation models within apps is pretty baller”

“Their Secure Enclave server bases AFM model is good, but not in the same class as gemini 3.5 flash”

“for small models they are good”

“If you're on macOS you can try the built in LLM which I think is similar in size.”

“It's perfectly capable of looking at images and classifying them at the level you'll need for this app.”

“that's actually pretty powerful for a use case like this”

“it's definitely enough to unlock features that rely on natural language understanding”

“seem great on paper until you see the 4k context window”

“Apple exposed the API in a fairly complete way. I integrated their local model in less than a day”

“surprisingly capable and has tool-calling support”

“they’re not made for conversation, so keep your expectations down in chat mode.”

“Subjectively, AFM isn't even close to Qwen. It's one of the weakest models I've used.”

“has a hard limitation — it can't process prompts with Spanish/non-English words, which is a dealbreaker”

“It forces you to essentially maintain two app experiences.”

“AFM is still incredibly premature and I'm experimenting with newer models that perform much better.”

“they did make it do something borderline useful while targeting a baseline of M1 Macs with 8 GB of RAM”

“just having it locally already installed is a huge selling point”

“came away impressed at the speed and accuracy of the LLM”

“Not Apple Foundation Models — unfortunately they're not capable enough (yet) for understanding content”

“Apples models are universally terrible in comparison to almost every open-weights model”

“There is no major opportunity for developers on Apple's platforms when they can just rug pull you as they please.”

“I also think the LLM is decent, and beyond its quality the scale of distribution is a big deal.”

“I've been somewhat surprised at the quality and reliability of Apple's built-in LLM and have only been limited by the logic I've built around it.”

“Their FoundationModels are actually pretty good at certain tasks”