1.4.5 - AI Slop Generation

Stable Diffusion Support	New Chat Design	Image Generation

Changelog:

🖼 Added image generation

• Support for stablediffusion.cpp, with model library and installation/compilation
• Support for Pollinations, OpenAI and Openrouter
• Support for image editing

💬 New chat redesign

🐞 Minor fixes:

• STDIO MCP servers are now supported on Flatpak version
• Better tool lazy loading: if a lazy loaded tool is called without tool search, ti will return its schema
• Fixed error window when MCP server is not working
• Arch-chan is now aware of if the system is running Nyarch or Arch

1.4.0 - Interfaces

Prebuilt llama.cpp
Commands Permissions	Reorderable Prompts
Better Text Rendering	Interfaces

Changelog:

🔗 Added Interfaces: alternative way external applications can interact with Newelle
✈️ Added Telegram support, you can now interact with Newelle using Telegram
🌐 Added OpenAI compatible API: most Newelle's functionalities are now accessible via an OpenAI compatible API, including TTS, STT, LLM and Embeddings
💬 Third party OpenAI compatible interfaces can control Newelle using commands with our v2 API
🖥 Added GUI API: Developers can now create third party interfaces that interact with Newelle using this API
👷 Support for directly download pre-compiled binaries for llama.cpp instead of compiling
🔐 Better Command permissions
🛡 When auto command execution is on, only safe commands are allowed
🚷 Dangerous commands are blocked
📐 Support for custom rules, including different paths
💬 Better Prompt Editing
📎 You can now reorder your prompts
🗂 You can now open your prompts in another window to edit them better
➕ You can now create your own extra prompts
📝 Better Font rendering
✍️ Improvements to text style defaults
⚙️ Additional settings to customize fonts
⛓️ Added support for markdown dividers
🛠 Added native tool calling: LLMs can now call tools natively according to their APIs, implemented in OpenAI handlers, Ollama and Gemini
☁️ Added full support for Ollama Cloud
🐠 Added support for Tinifish Websearch

🆕 Minor Improvements:

• Added a splashscreen on application start
• Fixed a bug that caused the application to take a long time to start
• Profile switch is now more responsive
• The chat will show the profile name that created the generation
• Added support for custom headers in OpenAI Handler
• Added the possiblity to specify an extra folder to search models file from in llama.cpp

Contributors

• @FrancescoCaracciolo

New Contributors

• @FabienDanieau made their first contribution in qwersyk#384

Full Changelog: qwersyk/Newelle@1.3.7...1.4.0

1.3.7

1. Fix terminal dialog not working on execute command tool
2. Commands with sudo will automatically run in dialog mode to avoid confusion
3. Added in the console prompt a notice when running with virtualization
4. Add a flag to indicate when a tool call requires interaction. If Newelle is unfocused when a tool that requires interaction is called, a notification will be sent
5. Added a new tool that allows the AI to ask questions to the user

6. Add mermaid rendering

7. Add an advanced mode to the read website tool so that it can fetch more content from a web page

1.3.6 Hotfix

Fix avatar loading and updating

1.3.5 - Agent Update

🛠 Improved Newelle Agent capabilities
📝 Added tools to explore and manage files (write, read,grep,glob, listdir)
🔐 Added files permissions: you can now specify for which files allow read,write or ask
✨ Added support for skills, Newelle can now activate skills, supporting huge libraries of existing ones
👥 Subagents: Newelle can now launch subagents to do subtasks. These will run with custom tools and skills
⏰ Scheduled Tasks: You can now schedule AI tasks to any moment
✔️ Added TODO List tool: Newelle can now keep track of complex tasks completion using the new TODO tool
⚙️ Tool lazy loading: choose which tools to lazy load to make context small and load their schemas only when needed
📄 Context Manager: Newelle will now dinamically manage the context in order to keep it in the limits
🔒 Added support for Oauth login for MCP tools
🗂 Added chat folders: organize your chats into folders
💻 Added commands: extensions can now add commands to Newelle

1.3.0

🗯 Multiple chats: you can now open and use multiple chats at the same time
🎤 Added wakeword support: you can now awake Newelle with a wakeword
📞 Added call mode: you can now call your models in real time
🔊 Streaming TTS support: TTS now answers faster for EdgeTTS, OpenAI Handlers and Kokoro
🎙 You can now compile Whisper.CPP with hardware acceleration
➕ New handlers: EdgeTTS, Llama.cpp (Embedding)
💾 Added "Agentic Memory", a new memory system that uses tools and new technologies in order to create a long term memory
🛠 Memory and RAG handlers can now expose tools, added tools to create TTS and STT
👁 Added vision support for Llama.cpp

1.2.5 - Chat branching and instant message rendering

What's Changed

⚡️ Implemented live message rendering while streaming
💬 New chat history design
🌳 Implemented chat branching from any message

Minor Improvements:

• Hide virtualization setting for non flatpak
• Add option to run system llama.cpp
• Improve chat name generation

Fixes:

• Improve stability
• Add a cancel button on llama.cpp install
• Implement native tool calling for openai and ollama
• Fix ollama not seeing pasted jpeg
• Fix google import error

2026-01-20T11-05-22-296Z_tagliato.mp4

1.2.0 - llama.cpp, model library and better document reading

⚡️ Add llama.cpp, with options to recompile it with any backend
📖 Implement a new model library for ollama / llama.cpp
🔎 Implement hybrid search, improving document reading

🐧 New tools to explore Arch wiki, AUR and official packages enabled by default
🗑 Removed smart prompts

💻 Add command execution tool
🗂 Add tool groups
🔗 Improve MCP server adding, supporting also STDIO for non flatpak
📝 Add semantic memory handler
📤 Add ability to import/export chats
📁 Add custom folders to the RAG index
ℹ️ Improved message information menu, showing the token count and token speed

Minor Improvements:

• Newelle is now easier to package for non flatpak
• Switch to Model2Vec as default embedding
• Add a spinner while the model is loading
• Improve application responsiveness
• Implement lazy loading for messages
• Option to hide history on launch
• Add search in file explorer
• Switch to ddgs for websearch
• Add tools to show video/images
• Better message generation stopping
• Add more options to the websearch tool
• Add keyboard shortcut to stop message generation
• Add tool settings to profiles
• Add an option for parallel tool execution
• You can now delete and edit console/tool messages

Fixes:

• Fix issues with openai handlers as secondary llma
• Fix ddgs not streaming
• Fix some random crashes
• Fix issues with automatic STT

1.1.0 - VRM and MCP support

• Added full support for VRM 3D Models, with expressions, animations and lipsync
• Added support for MCP (Model Context Protocol) and tools. Now you can import the tools from more than 10'000 servers
• Extensions can now easily add tools
• You can now add an LLM model to your favourites in the model popup
• You can now start/stop recording and stop TTS using commands (that can be bind to keyboard shortcut based on your DE)

1.0.1 - Runtime Update

• Update runtime
• Fix edge tts not working
• Sync with Newelle