Msty Studio Changelog

Quick Commands🔗

Studio Desktop 2.0.0-alpha.11

Features

• Quick Commands allow you to quickly access and run conversation commands. Use the back slash \ in the conversation input to access the Quick Commands menu.
• Knowledge Stacks now has a Msty Ignore setting at the folder level. This allows you to configure folders and file types to ignore when composing your Knowledge Stack. This is useful for excluding files that you don't want to include in your Knowledge Stack, such as temporary files or files that are not relevant to your Knowledge Stack. This applies to folder upload option; file upload option will still attempt to process the files you selected for composition.

Enhancements

• Keyboard Shortcuts save globally across all workspaces
• Icon selector is now searchable, offers a larger area to view more options, and allows keyboard navigation.
• Real-Time Data (RTD) for MLX models
• Various UI/UX improvements
• Knowledge Stacks allows remapping of embedding models
• List 'Default' embedding model in Knowledge Stack embedding model selector
• Made it easier to add custom models
• Improved visibility of Model presets
• Improved Toolbox UI/UX

Fixes

• Resolved an issue using online embedding models to compose Knowledge Stacks
• Fixed an issue with Persona Autoresponder
• Enable/disable of attachments in conversation from being included or excluded in context sent to model
• Resolved Knowledge Stack issue regarding an error when composing some files
• Fixed an issue with use Chunk Console when querying some imported Knowledge Stacks
• Resolved an issue when enabling a large number of Knowledge Stacks at once
• Fixed a UI issue with scrollbar and split menu

MLX Support, Keyboard Shortcuts, and GPT-5🔗

Studio Desktop 2.0.0-alpha.10

New in This Release

MLX Support is now available for Apple Silicon devices. You can now run MLX models on your Mac with Msty Studio Desktop. Please note that MLX support is experimental at this time and let us know if you run into any issues on our Discord. Studio Web is not supported at this time and support will depend on future developments by Apple and MLX project.

See the documentation for MLX Services to learn more.

Keyboard Shortcuts have been added for both web and desktop versions. You can now navigate and use Msty Studio more efficiently with keyboard shortcuts. Check the Keyboard Shortcuts documentation to learn more.

GPT-5 models are now available for use in Msty Studio. You can now use the latest GPT-5 models from OpenAI. We've included a few presets as well to get you going quickly.

Enhancements

• You can now edit model details such as name and purpose from Model Hub
• Your preference for submitting on enter or shift+enter is now saved across sessions
• Model list now automatically updated for OpenAI, Anthropic, Google Generative, and Mistral models
• Disable Web Security in Settings section - please use with caution, this will disable CORS
• Conversation titles can now be edited inline
• Local AI Services, including MLX, can be removed
• There is now a quick link from Model Hub to open AI Services Settings
• Right-click context menu is now available for projects, prompt folders, knowledge stacks, turnstiles, and personas
• Model Matchmaker stats updated

Fixes

• Fixed issue with Cerebras models not working properly
• Fixed Grounding Sources for Gemini Search preset
• Fixed issue with Personas not forgetting deselected models

Msty Studio Desktop Free + Knowledge Stacks 2.0 ⚡️⚡️ 🔗

Studio Desktop 2.0.0-alpha.9

New in This Release

Introducing Msty Studio Desktop — now free to use!
Studio Desktop is a major leap beyond the original Msty App v1.x, built for speed, flexibility, and power.

Included in the Free Version:

• Toolbox – Call MCP tools directly from conversations
• Knowledge Stacks 2.0 – A reimagined RAG experience
• Personas – Fully customizable AI assistants
• Much more – Attachments, split chats, prompt library, etc.

Check out the plan feature comparison matrix to see what features are included in the free version.

Meet Knowledge Stacks 2.0 – RAG, supercharged:

• Visualize chunks
• Built-in PII scrubbing
• Enhanced result views and analytics
• Knowledge Stack usage analytics and recommendations

To learn more, see the Knowledge Stacks documentation

Please remember, Msty Studio is still in Alpha. We don't anticipate any breaking changes with this release, but there may be unintended breaks all the same.

Also, these updates are for Msty Studio Desktop. Studio Web updates haven't been made with this release.

Plus:

• Amazon Bedrock support (for Aurum subscribers)

Check out the Bedrock documentation to learn more.

Enhancements

• Conversations now persist personas, models, and split chats across sessions
• Expand/collapse all Projects in one click
• Drag-and-drop organization for left Nav Bar, Chat Add-ons, Projects, Prompt Library, and more

Fixes

• Fixed an issue with editing system prompt after conversation has started

Msty App to Msty Studio Desktop Data Migration 🔗

2025-07-18

Studio Desktop 2.0.0-alpha.7

New Features

• Data migration from Msty App to Msty Studio Desktop is now available. Please see our Migration Guide for details.

Improvements

• UI/UX for model selection dropdowns when adding models for online providers - more options now display in the view
• Primary modals for Model Hub, Personas, Toolbox, etc can now be minimalized and will be remembered as a preference for that modal
• Projects lists refreshes automatically when a project is removed
• SearXNG endpoint now persists between sessions

Fixes

• 'X' button inactive for some drawers
• Ellipsis for split chats not working properly when there are 3+ split chats

Added App Info Settings 🔗

2025-07-14

Studio Desktop 2.0.0-alpha.6

New Features

Application Information is now available in Settings > General. From here, you can see what version of Msty Studio desktop you are running. You can also set Studio Desktop to auto-update as well as manually check for updates.

Additionally, you can locate data path, app logs, reset app settings, and more.

We've also brought 🎊 back to new app updates.

Improvements

• Improved error messages for models that do not support tools or thinking

Multiple new fixes and improvements hitting Msty Studio 🔗

2025-07-11

We've released a new version of Msty Studio Desktop (v2.0.0-alpha.4) and Studio Web containing multiple new features, improvements, and fixes.

For desktop, the updates will install in the background.

New Features

Improvements

Fixes

N/A - not applicable for the platform type

* Will be included in a future release

Announcing Msty Studio Desktop - now available in Alpha for Aurum subscribers 🔗

2025-06-30

Check out the full Announcement here.

Msty Studio is going to be unified across web and desktop.

If you are an Aurum subscriber, you can now download the Msty Studio Desktop app and start using it. Check out the Studio Desktop Docs for more details on how to install and use the app.

Real-Time Data (RTD) Enhancements 🔗

2025-06-26

New search providers for Real-Time Data (RTD) have been added to Msty Studio, including:

• Tavily
• Exa
• Google Search API

In addition to the new providers, we've also made some improvements to UX and Sidecar RTD results.

Msty Sidecar Beta Release and a few new enhancements 🔗

2025-06-23

Msty Sidecar Beta Release

Msty Sidecar is officially getting ready for a version 1.0 release. We have the beta version available, please check out this Discord message for install links.

New features and improvements in the Sidecar Beta include:

• Update Local AI from within the app - no need for us to push new updates or do it manually
• Added a .token file where you can add your own custom token under App Data folder
• Improved onboarding experience
• Improved tool dependencies discovering
• Set setup window to always be on top during onboarding
• Reduced installer size
• Improved RTD results
• Fixed a YouTube transcript reading issue

Let us know on Discord if you run into any issues or have any feedback.

Improvements and Fixes

• Added ability to exclude project context in the system prompt, this is now an opt-in feature
• Allow importing tool config that follows the same format of Tool Configuration in Msty
• Fixed an issue with think tag

Advanced Tool Calls and Gemini 2.5 Pro and Flash Models 🔗

2025-06-17

New Features

• Advanced Tool Calls - tool call usage now displays inline in the conversation, allowing you to see the tool call and its response. You can also select View Tool Usage to drill into tool calls from any conversation.
• Gemini 2.5 Pro and Flash models are now available for use in Msty Studio

Fixes

• Latest tool call overwrite the previous
• Attachment tab not correctly displaying uploaded files
• Tooltips not hiding automatically in some instances

House Keeping 🔗

2025-06-05

Fixes

• System prompt takes the full height of the view
• Local models displaying true/false for purpose tags instead of the purpose's label
• Personas overflow menu does not always align with ellipsis
• Active download tooltip arrow doesn't properly align
• Selecting a persona during a conversation scrolls window back to top
• uv => uvx
• Split chats being created when exiting vapor mode
• Cannot delete multiple conversations using the floating UI

Auto Responder Persona and Context Isolation 🔗

2025-06-02

New Features

The new Auto Responder Persona watches the conversation you are having with a model and responds to the conversation according to your directions for the persona.

With Auto Responder, you can setup a:

• Fact Checker Persona that fact-checks the conversation
• Deeper research Persona that looks further into a statement being made by a model
• Devil's advocate Persona that argues the counterpoints
• And more!

Also new and related to Auto Responder is the ability to isolate any message from the conversation's context without removing the message. This allows you to really dial in on specific LLM responses in the conversation thread.

Check out the Auto Responder Docs to learn more.

Watch the Youtube demo of Auto Responder and Context Isolation to see it all in action.

Improvements and Fixes

• Set the context message count for a Persona
• Moved scrollbar to the edge when split widths are not full width
• Fixed streaming message indicator not being displayed next to the title

Image Generation and Prompts Library 🔗

2025-05-30

New Features

• Generate images using supported models such as Gemini Flash and GPT Image. Check out our Image Generation Docs for more details.
• Prompts Library lets you create, organize, and manage prompts that you can use throughout Misty
• Hide messages from conversations without taking them out of the context history

New Gemini and Anthropic models plus UI/UX improvements 🔗

2025-05-28

Improvements

• Added new Gemini and Anthropic models
• Updated Model Matchmaker
• Improved UX for Live Context
• Made some additional UI/UX tweaks

Environment Variables and Shiftcuts 🔗

2025-05-09

New Features

• Environment variables are now available in Msty Studio. You can manage variables for multiple environments and quickly switch between them.
• Shiftcut mode - hit shift twice quickly followed by one of the following keys to activate shortcuts for that specific task
  • t - create a new tab
  • n - create a new conversation
  • 1 - toggle sidebar
  • e - show Quick Select Environment Variable picker

Azure OpenAI Integration 🔗

2025-05-07

New Features

Msty Studio now supports integration with Azure OpenAI. Check out the link to the docs for a setup guide that takes you through setting up both in Azure and in Msty.

Backup and restore 🔗

2025-05-06

Improvements

Msty Studio now prompts for backups every 24 hours. You can change the frequency in the settings section.

Also, you can now export backups and restore them from the settings section.

Context shields in Turnstile 🔗

2025-05-02

New Features

You can now add context shields to your Turnstile conversations - with or without a summary.

You can use --cs-- and --css-- shortcodes to add context shields to your Turnstile conversations.

Improvements

Left rail icon tweaks.

Fixes

• Broken scroll when editing raw markdown
• Can't add, continue, or repeate shortcodes in Turnstile
• Tooltips that didn't want to close

Qwen3 models available for local download 🔗

2025-04-28

New Features

• Qwen3 models now available
• Preserve vapor chat

Bug swatting 🔗

2025-04-27

New Features

Hide a model's provider from model selection list

Fixes

• Fixed a few issues with Provider Options
• Cannot export params preset if no params is set even if advanced and/ or provider options set
• Open Router doesn't recognize Msty Studio
• Sidecar can't connect if the url contains a path such as https://example.com/sidecar - this is useful for those running Tailscale

Gemini 2.5 Flash Preview and advanced options and configs 🔗

2025-04-25

New Features

• Provider Options as an advanced config to pass a provider specific options such as thinking budget, reasoning effort, etc - these are for advanced use cases
• Pass thinkingBudget and reasoningEffort as a model params without worrying about provider options
• Gemini 2.5 Flash Preview now available
• 2 new model params presets

Ability to delete multiple conversations 🔗

2025-04-22

New Features

Ability to delete multiple conversations within a folder (use Shift or meta/win key for selecting multiple)

Improvements

Turnstile QoL improvements:

• Show model being used
• Open panel by default
• Use starred model by default

Fixes

• License key doesn't need to be activated in the app first
• Model purpose tag showing true instead of purpose value

o3 and o4-mini support 🔗

2025-04-16

New Features

• ChatGPT o3 and o4-mini support now available

SearXNG support and UX improvements 🔗

2025-04-15

New Features

• Support for SearXNG search engine
• Re-name models to be more intuitive for you
• Assign categories to models for easy recognition
• Edit conversation titles

Search Grounding, GPT 4.1, and Bug Fixes 🔗

2025-04-14

New Features

• Search Grounding support for Gemini and Perplexity models
• Show sources for search grounding supported models
• Params preset for search grounding
• Support for the new GPT 4.1 models

Fixes

• Python being falsely flagged as being not-detected
• System prompt input box height