Grok Voice Transcribe 2.0 Added to Speech-to-Text API

The developer release notes track modifications and new features across the platform's API services. In this update, the documentation records the rollout of a second-generation audio transcription model. While the new model identifier can now be specified in developer requests, the original first-generation transcribe model remains the standard default for incoming calls. Additionally, the entry drops recent feature tags previously placed on custom voice options. For developers building on this platform, the change means they can deliberately opt in to the newer transcription engine by updating their model parameters. Workloads relying on default configurations will continue operating on the first version without disruption, avoiding unexpected breaking changes. However, the release notes do not describe the underlying differences between the two model versions. The documentation omits whether the newer model carries different usage pricing, distinct rate limits, or altered quota requirements. It also does not specify whether the newer version introduces improvements to transcription latency, recognition accuracy, or audio language support.