hyperframes Releases

This release focuses on bug fixes across documentation, CLI, player, studio, and engine components. It addresses audio playback issues, improves player stability, and updates documentation schemas and tooling.

Apr 18, 2026

This release primarily resolves audio playback issues in the player and migrates documentation assets to a new CDN.

Apr 17, 2026

Minimal release notes: "**Full Changelog**: https://github.com/mukul975/cve-mcp-server/commits/v0.1.0"

Apr 14, 2026

This release introduces hooks support, a new `/clear` command to reset chat history without losing runtime state, and UI improvements like a code copy button. It also adds Catalan language support and fixes critical conversation persistence bugs.

Mar 30, 2026

This release adds a planning agent, a real-time task list view, and a slash command menu. It also enhances repository management by allowing Git repo attachment changes and adds support for several new AI models and Bitbucket datacenter.

Mar 11, 2026

This release adds support for the MiniMax-M2.5 model and fixes a bug where resumed conversations would incorrectly display an error status during startup.

Feb 18, 2026

This release introduces a new LoRA-based abliteration engine with 4-bit quantization support, implements Magnitude-Preserving Orthogonal Ablation, and adds features for saving and resuming optimization progress. It also includes improvements for GPU enumeration, memory management, and broad support for VL models.

Feb 14, 2026

This release introduces significant hardware support including Apple Silicon (MPS) and IBM Granite MoE, alongside multi-GPU capabilities. It also enhances model compatibility, reduces memory usage, and adds notebook support for easier experimentation.

Dec 10, 2025

Nov 16, 2025

Introduces a new graphical user interface (GUI) for fine-tuning TTS outputs, allowing users to precisely control pitch and duration for individual phonemes and swap voices across 7000+ languages.

Oct 7, 2024

A significant quality improvement release focusing on the vocoder, which is now trained end-to-end on synthetic spectrograms to bridge the gap with natural audio, resulting in better sound quality without sacrificing speed.

Sep 22, 2024

Enhances the universal TTS model by introducing stochastic prosody prediction and adding more IPA modifiers and languages to the pretraining pipeline.

Jul 25, 2024

A major milestone release enabling Text-to-Speech in over 7000 languages, achieved by extrapolating from a 462-language model and adding watermarking for misuse prevention.

Jun 10, 2024

Introduces emotion control to the TTS model, allowing users to condition the synthesis on emotional prompts and transfer emotions from training data to generated speech.

Jun 10, 2024