We're excited to share our latest updates to the EdgeRunner platform!

You can now refine prompts, highlight text for quick actions, export entire chats, and experience instant text-to-speech streaming.

Combined with faster performance across document and prompt processing, improved history organization, and smarter search, the chat experience is now more responsive and easier than ever before.

The update is now available in the EdgeRunner DoW Beta here.

_________________________________________________________

New Features:

• Editable User Prompts – Users can now edit their prompts directly in the chat. The assistant automatically regenerates a new response based on the updated input.
• Contextual Highlight Actions – Users can highlight any portion of the assistant’s response to access quick actions such as Explain or Translate, enabling deeper exploration of responses.
• Chat Export – Entire chat histories, including all prompts and responses, can now be saved in formats like PDF for easy sharing or recordkeeping.
• Chat Title Renaming – Chat titles can now be renamed within the History Section, helping users stay organized and easily identify conversations.
• Copy Prompts Instantly – A new “Copy” icon allows users to quickly copy their prompts to the clipboard, displaying a confirmation toast for successful completion.
• Search Chat History – The new search feature enables users to find previous conversations quickly using keyword-based search within the Chat Section.
• Document Title Search – Users can now search for documents by entering all or part of the title within the Chat Section, making it easier to locate uploaded files.
• File Collections & Management – Introduces streamlined tools for managing and grouping uploaded documents into organized collections for quick access.
• Reasoning Model (CoT) – Adds support for the new EdgeRunner 20B Reasoning Model, based on gpt-oss-20b, enhancing complex, multi-step reasoning and structured analytical capabilities. This is available for the EdgeRunner-Medium version.
• LoRA Hot-Swap – Introduces support for modular LoRA adapters on a single base model, allowing targeted personalization without retraining or replacing the model.
• Citations – When performing Retrieval-Augmented Generation (RAG), a citation chip now appears showing the retrieved content, its document source, page number, and author for full transparency.
• Text-to-Speech (TTS) Streaming – The new “Speak Aloud” feature now streams audio playback for assistant responses instantly, providing a seamless listening experience.
• Speech-to-Text (STT) – Users can now click the microphone button to speak prompts directly. Speech is converted to text instantly, streamlining hands-free interactions.

Improvements:

• Performance Upgrades – Major backend optimizations deliver faster performance across all key workflows:
  • Document processing is now 45% faster, allowing files to load and process more quickly.
  • Prompt processing is now 230% faster, enabling responses to generate more than twice as fast.
  • Token generation is now 10% faster, creating smoother streaming and quicker completions
  • Text-to-speech performance has improved by 877%, allowing responses with over 5,000 characters to play back almost instantly.
• Inference Overhaul – Model inference speed has been improved dramatically, achieving up to 452% faster performance compared to Ollama with GPT-OSS, resulting in near-instant response times.
• Model Retention on Switch – When switching between models, the current conversation and any attached documents now remain active and visible, preserving context.
• Auto-Reorder Recent Chats – When a user updates an existing chat, it automatically moves to the top of the History list, keeping active conversations easily accessible.
• Drag-and-Drop Uploads – Users can upload documents effortlessly by dragging them directly into the chat window.
• Software Cleanup – All proprietary dependencies have been removed, making the system more flexible, secure, and risk-free for open deployment.
• General Enhancements – Includes numerous performance optimizations, user interface refinements, and workflow improvements to ensure a smoother and more consistent experience across the platform.

Bug Fixes:

• Fixes and Stability Updates – Minor issues have been addressed, and overall application reliability has been improved to ensure a more stable user experience.