Apple has made major adjustments to its Siri development strategy under the leadership of Craig Federighi. Apple engineers are now allowed to use third-party large language models (LLMs) to develop Siri features for the first time, breaking the previous restriction of only being able to use internal models.

In the past, Apple strictly required engineers to use only self-developed large language models to develop Siri functions. In 2023, management explicitly prohibited the integration of external models such as OpenAI ChatGPT into the final product.

However, compared with OpenAI's technology, Apple's self-developed model has a performance gap, and the software team has always been restricted during the development process and unable to create powerful AI functions. During internal testing, although engineers have extensively tested the OpenAI model, they were unable to apply it in practice due to policy restrictions.

After Craig Federighi took over the Siri project, the situation changed rapidly. He adjusted the development strategy, emphasizing that even if external AI models were introduced, the best AI functions should be created at all costs. This policy broke Apple's long-standing closed development tradition.

According to people familiar with the matter, Federighi's decision was based on user demand - users are more concerned about the actual performance of Siri's functions rather than the source of the technology behind it. Now, Apple engineers can freely choose the most suitable open source LLMs to inject more powerful capabilities into Siri.

According to IT Home, this move is considered a key transformation for Apple in the field of AI. Previously, Siri was criticized for its lack of functions, and opening up third-party models is expected to significantly improve its competitiveness.