BREAKING
+++ OpenAI launches Ultrafast: GPT-5.6 Sol at up to 750 tokens/second +++ Apple trains its own AI model for China — with Alibaba +++ Zhipu's GLM-5.3 beats Mythos 5 and GPT-5.6 Sol on CyberGym +++ DeepSeek hikes API prices — cache hits more than 10x more expensive +++ OpenAI run rate tops $40 billion ahead of IPO +++ Gemini 3.7 Flash: big coding leap just three weeks after its predecessor ++++++ OpenAI launches Ultrafast: GPT-5.6 Sol at up to 750 tokens/second +++ Apple trains its own AI model for China — with Alibaba +++ Zhipu's GLM-5.3 beats Mythos 5 and GPT-5.6 Sol on CyberGym +++ DeepSeek hikes API prices — cache hits more than 10x more expensive +++ OpenAI run rate tops $40 billion ahead of IPO +++ Gemini 3.7 Flash: big coding leap just three weeks after its predecessor +++
Updated 08:00
AI IN LIFE AI IN LIFENEWS
DAILY
Business

Apple Trains Its Own AI Model for China — With Alibaba's Help

Apple has built a proprietary LLM for the Chinese market, Reuters reports — making it the first foreign company approved to offer its own AI model there.

Apple Trains Its Own AI Model for China — With Alibaba's Help

Illustration · AI-generated (AI IN LIFE)

At a glance

  • First proprietary AI model by a foreign company approved in China
  • Alibaba supported the model's development
  • Alibaba's Qwen is set to serve as an optional extension
  • AI feature rollout in China expected in the coming months
  • Reported by Reuters on August 14, 2026

Apple has trained its own large language model for the Chinese market to power Apple Intelligence features there, Reuters reported on August 14, citing people familiar with the matter. Alibaba supported the development — and regulatory approval reportedly makes Apple the first foreign company allowed to offer a proprietary AI model in China.

The move marks a strategic shift. Apple had widely been expected to rely primarily on models from local providers in China. Instead, the company is pursuing a dual-track strategy: its own China-trained model as the foundation, supplemented by Alibaba's Qwen as an optional extension, similar to the ChatGPT integration in other markets.

The backdrop is fierce competition with Huawei and other domestic smartphone makers, whose devices have long shipped with locally developed AI features. Apple repeatedly had to delay Apple Intelligence in China because generative AI services must complete a government registration process there.

The rollout of AI features in China is now expected “in the coming months.” In a curious footnote, guidance on connecting Siri to Qwen briefly appeared on Apple's support pages before being withdrawn within a day.

For the industry, the case illustrates how geopolitical fragmentation is forcing separate technology stacks: competing in the Chinese market requires locally registered models, local partners, and local infrastructure — even for the world's most valuable company.

◈ AI-GENERATED REPORT · SOURCES LINKED

FAQ

Why is Apple building its own model for China?

Generative AI services require government registration in China. A proprietary, locally approved model gives Apple more control than relying solely on Chinese third-party providers.

What is Alibaba's role?

Alibaba supported development, and its Qwen model is expected to be integrated as an optional extension — comparable to the ChatGPT integration elsewhere.

When will Apple Intelligence launch in China?

According to the Reuters report, the rollout is expected in the coming months; Apple has not announced an official date.