v0.8.143 v0.8.144

Name: hyperframes
Author: heygen-com

low risk

heygen-com/hyperframesOct 8, 2026 → Oct 10, 20261 day between releases

What's New

CLI gains transcription export and Windows stability improves.

New Features

CLI: `transcribe --to` writes directly to SRT or VTT formats.
CLI: Models list indicates which speech models support transcription.
Registry: Added 30 new WebGPU shader backgrounds.
Engine: Performance optimized for audio duration decoding.
Core: Previews now preserve all layers and stop looping past end.

Improvements

CLI reliability improved on Windows, including fixes for rendering cancellations and unzip errors.
Engine performance optimized by recovering audio duration via final frame decoding.
Preview stability improved to prevent small previews from dropping layers.

Bug Fixes

Core: Stop small previews dropping layers on heavy compositions.
Studio: Timeline viewport gate gives consistent verdicts.
Core: Fixed data script handling.
General: Previews steadier and CLI reliable on Windows.
Engine: Recover audio duration by decoding final frame.

Upgrade Notes

The `transcribe` command now supports `--to` for direct SRT or VTT export. Windows users will experience improved CLI stability and preview rendering. No configuration changes are required.

The Two Releases

v0.8.143

Oct 8, 2026· 13 days ago

This release significantly improves the performance of distributed renders for long videos by changing how frames are extracted, reducing planning time on AWS Lambda from over 6 minutes to roughly 20 seconds. It also includes extensive bug fixes for WebVTT parsing, CLI commands, and SDK edge cases.

Full release notes

v0.8.144

Oct 10, 2026· 12 days ago

This release focuses on improving preview stability, particularly on Windows, and enhances CLI reliability with new export options. It also introduces new WebGPU shader backgrounds and optimizes engine performance.

Full release notes