OS-native model API
Operating systems increasingly ship built-in, benchmarkable on-device model capabilities (speech, vision, language) as a stable system API that apps call instead of bundling and shipping their own weights.
grounded in: HN front page (2026-07-14): Apple's on-device SpeechAnalyzer speech-to-text API benchmarked head-to-head against Whisper — an OS-provided model API vs. a bundled model.
Connected concepts
On-device speech recognition, Platform-native inference, Provider abstraction, Streaming ASR
Explore it live in the knowledge graph →