This article doesn’t have any sources, and all of the sites talking about this are pretty unreliable. I’ll believe it’s real when the reliable rumor sites pick it up.
That said, Apple Intelligence has three tiers of prompt processing. Apple’s private on-device model, Apple’s private cloud model for more complex prompts, and external model integration for the most complex prompts processing. The latter is model agnostic.
Apple launched with ChatGPT as an optional integration, but in theory, it could plug into whatever. Gemini, DeepSeek, etc.
Apple will probably just swap GPT for DeepSeak in China, just like they swap Google for Baidu with search.
Exactly. This is a non-story and they already have a built in integrations system for escalating / customizing requests.
I could see them maybe using a custom version of Deepseek for that second tier. Could be a boon for them if the inference costs are a lot lower.
Smella of sus
