camel Releases

Minimal release notes: "**Full Changelog**: https://github.com/cortezaproject/corteza/compare/2024.9.5...2024.9.6"

Nov 19, 2025

Nov 15, 2025

Introduces DSQ (DeepSeek Quantization) to reduce VRAM usage by patching linear layers, supporting DeepSeek-OCR and PaddleOCR-VL with Q4_K, Q6_K, and Q8_0 precision tiers.

Nov 14, 2025

Adds AiHubMix and SiliconFlow model support, along with AWS Bedrock async improvements and Gmail toolkit integration fixes.

Nov 13, 2025

Enhances the Papercut SMTP UI with dark theme support, IP allowlist filtering, and a dynamic property editor, while upgrading to .NET 9.0.

Nov 11, 2025

Addresses a critical memory issue on Blackwell GPUs where the RoPE cache was incorrectly expanding to massive positions, causing crashes.

Nov 10, 2025

Adds the PaddleOCR-VL backend, implements lazy loading for faster startup, and introduces comprehensive sampling controls for the server and CLI.

Nov 9, 2025

Improves voice activity detection quality on edge cases including unusual voices, child voices, and muted speech through a reworked training paradigm.

Nov 6, 2025

Publishes a fix for PyTorch 2.9 compatibility and refines tuning utilities by removing ipdb and fixing optimizer gradients.

Nov 5, 2025

Fixes critical email forwarding rules that were failing due to TaskCanceledException and corrects SSL/TLS connection modes.

Nov 1, 2025

Adds comprehensive UI improvements including search functionality, zoom controls, attachment display, and full SMTP authentication/TLS/STARTTLS support.

Oct 28, 2025

This release focuses on stability by fixing critical bugs related to Transformer model loading, model selection state management, and the prevention of empty message sending.

Oct 25, 2025

This update improves reliability by fixing SearxNG timeout issues and introduces local storage persistence for model selection values and run commands.

Oct 23, 2025

This release focuses on fixing cross-system encoding issues, adding contributor badges to the repository documentation, and reverting a previous tool call caching implementation due to issues.

Oct 15, 2025

A minor maintenance release that updates model documentation and adjusts the default configuration for tool output caching.

Oct 12, 2025

A substantial release introducing AMD platform support, significant new toolkits (WeChat, DingTalk, Google Veo, Slack, Microsandbox), and workflow enhancements for the workforce system.

Oct 11, 2025

This release modernizes the build system, adds Windows ARM64 support, and enhances Docker configurations with non-privileged ports while fixing service installation issues.

Oct 10, 2025

This is a minor release with no detailed changes provided in the snippet, only a link to the full changelog.

Oct 6, 2025

This release ensures backward compatibility with older versions of Gradio Lite and prepares the project for the new version.

Sep 17, 2025

This release includes various bug fixes for the async client, input handling, and UI rendering, alongside dependency updates.

Sep 2, 2025