- Share link now correctly includes subpaths.
- Synonyms display corrected for read-only assistants.
- OAuth first-login visual issues resolved.
- Nav Drawer branding fixed.
- Facets dropdown no longer disappears after selection.
- Chatbot "go to last message" arrow no longer overlaps other elements.
- Visual and contrast adjustments for dark mode.
- Tooltips added for compact mode navigation.
- Links in generated answers now open in new tabs.

- Additional metadata per file exposed on the file view.
- File status indicators and indexing progress displayed in clip view.
- Cleaner scrolling behaviour across panels and containers.
- New toggle option for recurrent questions in Search Tasks.
- Improved padding and layout for Search Tasks.
- More stable text rendering inside the chatbot.
- Metadata values now visible in file view.
- Better login-page layout and logo handling.
- More robust indexing, refresh tracking, and aggregation visibility.
- Assistant sharing now respects organisation-level access rules.
- Cleaner integration for audio encoding and transcription workflows.

- Prevent multiple messages being sent on Enter.
- Shared assistants: “UseOnly” hides download actions.
- CSS fixes on Search bar.
- Lang not shown before selection (EN) fixed
- Accent letters in assistant names allowed.
- After assistant rename, old name no longer lingers in Chatbot Conversation.
- Exported PDF uses conversation name.

- DateFilter colors fixed in dark mode.
- Connectors page: align body to top; consistent card sizes/heights.
- Increased timeout before naming new conversation.
- OAuth username collision fixed.

- Save original uploaded filenames.
- Fix empty conversation titles.
- Restore broken conversation title editing.
- Correct empty AI assistant list display.
- Fix overflow for long AI assistant titles.
- Ensure copying LLM answer includes sources.
- Catch model built-in guardrail errors.
- Adjust robots.txt enforcement (less restrictive).
- Prevent scraping of non-subpages.
- Fix login cache bug causing 404s.
- Resolve denied login page issue.
- Feedback box resized for better usability.
- Removed duplicate scrollbars.
- On-Premise Safe Text hotfix applied.
- Show file statistics in shared conversations.
- Ensure failure reasons stored for indexed files.
- Handle rate limit exceeded in iframe gracefully.
- Fix API key for /api/v1/chat/completions endpoint.,

- Logo included in conversation export PDF.
- Multilingual character support in conversation export PDF.
- Checkbox required when switching LLM from private to public.
- Allow slashes in AI assistant names.
- AI assistant search filters exposed directly in chat interface.
- Display release version in the UI.

- Clearer save button state for local changes, better explanation of prompt slots, and clearer configuration names/descriptions.
- Fixed various CSS bugs for a smoother experience.

- LLM & Embedding Provider Customisation: Choose the models and providers that fit your needs.
- Banner for LLM Types: Clear visibility whether you’re using QAnswer’s own models (on-premise) or external ones.
- Clip on Assistants: Upload data directly in assistant task chats, synced with the global chatbot clip feature.

- Improved Facets: Auto-generated filters are now clearer and easier to use.
- Auto Date Filtering: Search results can now be narrowed down by date automatically.
- Facets on Search: Enhanced filters to help you quickly find what you need.
- Cited Sources with Dates & Icons: Sources now include refresh dates (optional) and helpful icons.
- Query expansion: Expand the original query with additional terms or phrases.

- Confluence Integration Page: Connect and search your Confluence spaces →- Publication Office Connector: Unified with the new facets experience.
- EConsilium Connector: Another powerful integration for your workflow.
- Docling-as-a-Service: Seamless access to external docling service.
- SharePoint Search by Item Types: Find the right kind of content faster.

- Quota Exceeded Redirect: Users are now guided to their page when limits are reached.
- Access-PDF Experience: New loader and better disabling for a smoother experience.
- Role-based Access: Paths are now restricted based on user roles.
- General CSS Fixes: Cleaner design, reduced scrolling issues, and a more polished look.

- Claude 3.7 Sonnet Integration: Fully hosted in the EU and available alongside QAnswer LLM, GPT-4o Mini, and Mistral Medium.

- Human-in-the-Loop Support: Seamless switch from AI to human agents within the same chat interface with instant notifications.

- Smarter Report Generation: Reusable templates, import of Word/Markdown/HTML formats, and enforced structure for standardized reports.
- Smarter Acronym Handling: Acronym definitions injected into prompts for more accurate query understanding and response accuracy.