Apple's upcoming iOS 26 and macOS Tahoe are expected to include significant AI improvements, including a speech API that outperforms OpenAI's Whisper in transcription speed and quality, according to hands-on testing. A test involving a 34-minute podcast episode showed Apple's transcription model being 55% faster than OpenAI's best model. The new AI models will be free for developers, potentially improving the market for audio transcription when the updates are released.
Apple is reportedly making substantial advancements in artificial intelligence with its upcoming iOS 26 and macOS Tahoe operating systems, notably featuring a new speech API for transcription that demonstrates superior performance over established competitors like OpenAI's Whisper. Hands-on testing by MacStories, using a custom utility named Yap, revealed Apple's model transcribed a 34-minute, 7GB 4K video podcast episode in just 45 seconds, a processing speed 55% faster than OpenAI's MacWhisper (Large V3 Turbo), which took 1 minute and 41 seconds for the same task, while maintaining comparable transcription accuracy. Although the Apple model exhibited minor imperfections, such as occasional difficulties with last names and specific terms like "AppStories," its efficiency is a significant development. Crucially, Apple intends to offer these advanced AI models to developers free of charge, a move poised to invigorate the audio transcription market and foster innovation within Apple's app ecosystem once the new OS versions and Xcode 26 are publicly released. This development underscores Apple's commitment to enhancing its AI capabilities and integrating them deeply into its platforms, potentially strengthening its competitive edge in the technology sector.
AI-powered research, real-time alerts, and portfolio analytics for institutional investors.
Request a DemoOverall Sentiment
strongly positive
Sentiment Score
0.75
Ticker Sentiment