This page summarizes the 2026.12 release. Click the links below to navigate to the relevant sections.

📌 Release Overview

• 2026.12

• Unique hosted: March 18, 2026

• Self hosted: TBD

Chat Platform

🚀 What's New (Business/End-User Focus)

Admin Bing Grounded WebSearch graduates to GA: As part of Web Search Search Engine Setup and Configuration graduates to General Availability now, providing a viable alternative to the Search Engine Setup and Configuration which Google™ is discontinuing.

Admin MCP Server Usage tracking in Product Metrics: MCP server usage tracking is now included in the product metrics analytics data.

Chat Messages UI Refresh

This release introduces a refreshed way user messages and assistant answers are displayed in the chat thread. The new answer formatting is available in Beta , opt-in and disabled by default. It can be enabled on request via feature flag for selected users or environments.

User BETA Cleaner answer layout: User messages use a bubble style on the right; assistant messages retain a flat style on the left for a clear question-and-answer distinction. Assistant message avatars are no longer shown, making threads easier to read, especially on mobile.

User BETA Sources and content with the answer: Files and sources linked to an answer are shown as clickable badges directly under that answer’s text instead of as separate blocks between messages.

User BETA Unified message actions: Copy, feedback (thumbs up/down), references, and debug sit in a single action row for assistant messages. References and debug open in a side panel so you can keep reading the thread while viewing references or debug info. On user messages, actions appear on hover to keep the default view minimal.

User BETA Tables with copy: Tables in assistant messages use an updated design with a copy button in the lower-right of the table (visible on hover on desktop, always visible on mobile) so you can copy table data quickly.

User BETA Updated suggested follow-up questions: Follow-up questions use a clearer design with a list-style layout so suggested next questions are easier to scan and tap.

User BETA Floating hallucination indicator: The previous Answer Assessment button is removed. If no hallucination is detected, nothing is shown. When the model flags potential inaccuracies (moderate or severe hallucination), a floating button appears in the lower right (yellow for moderate, red for severe). Hover over the button to view the reason for the hallucination. It stays visible while you scroll so you can still see and act on the warning.

References in Chat UI Refresh

This release introduces a refreshed way references are displayed in the chat thread. The new references visualisations are going to Beta state, opt-in and disabled by default. They can be enabled on request via feature flag for selected users or environments, these will be moved to an env variable upon GA launch. Only one at a time can be active.

User BETA More references in line: Up to three references are shown in line, clicking the last button (+n) will expand and show all references in line (if more than 3 are present). A collapse button will appear at the end.

User BETA Numeric references in line: Up to five numeric references are shown in line, clicking the last button (+n) will expand and show all numeric references in line (if more than 5 are present). A collapse button will appear at the end. The same numbering is also shown in the references panel for each reference.

🐞 Bug Fixes

User Prompt Update Notification: A success toast now appears when a user prompt description is updated.

User References & Debug Button State: References and Debug buttons no longer incorrectly show an active state when the side panel is open.

User Chat Input Height: Fixed a layout issue where the chat input could miscalculate its height during split-pane resizing.

RAG

🚀 What's New (Business/End-User Focus)

Admin Knowledge Base Product Metrics Tracking: Introduced product metrics tracking for the Knowledge Base to measure user engagement and activity. Events such as Knowledge Base visits, document uploads, searches, and scope creation are now captured and aggregated into daily, weekly, and monthly active user metrics per company, enabling better usage insights and reporting.

🔧 Improvements

Admin Configurable Excel and CSV Ingestion Limits: Excel and CSV ingestion limits (maxRows and maxCols) can now be configured directly through the Configure File Ingestions settings. Previously, these limits could only be set via the SetScopeProperties API. This update allows administrators to manage spreadsheet ingestion limits more easily through configuration, improving flexibility and reducing the need for manual API calls.

Agentic Table

🐞 Bug Fixes

User Delete button acts as link when disabled during agent run: When the table agent is running, the delete button in the view-files area is disabled and no longer opens the file on click; a tooltip is shown instead.

User References in answer library side panel use correct colors: References in the Answer Library side panel use the same color coding as the References column (KB Answers vs Library Answers), so the distinction is consistent when opening a reference.

User Mac: Ctrl + trackpad click on cell does not show cell history context menu: On macOS, Ctrl+click (secondary click) on a table cell now correctly opens the context menu so "View cell history" is available when using a trackpad.

Chat Widget

🔧 Improvements

Admin New Env: APP_ROOT: /srv/runtime

• The chat-widget container now expects APP_ROOT to point at a writable directory (in QA/UAT1 : /srv/runtime).
  With a read-only root filesystem, the entrypoint copies the app from /app into APP_ROOT and then injects runtime config (e.g. window.uniqueRC) into the files there.
  Ensure the deployment sets APP_ROOT (e.g. to /srv/runtime) and that this path is backed by a writable volume (e.g. the existing runtime-dir emptyDir).

Unique AI

🚀 What's New (Business/End-User Focus)

User Code Execution Significantly faster download speed for generated files. Code Execution tutorial now available on Unique Academy.

User Deep Research Deep Research reports can now be exported.

Admin Bing Grounded Web Search – GA The grounding Bing search has transitioned from Beta to General Availability. This release encompasses comprehensive infrastructure support, including Foundry integration, Bing resource provisioning, and automated agent creation within the Foundry environment.

To enable this capability for your account, kindly contact our Customer Support team, who will be happy to assist with the activation process.

🔧 Improvements

User Code interpreter postprocessor improvements Enhanced code interpreter to display PNG images directly in chat while keeping download links for other files like spreadsheets and PDFs. Images now show with clear titles and include the code that created them. This feature can be safely enabled or disabled without affecting existing functionality using the feature flag FEATURE_FLAG_ENABLE_CODE_EXECUTION_FENCE_UN_17972.

Admin File Upload Write URL override Added an environment variable to override the writeUrl for uploaded files supporting custom deployments config

🐞 Bug Fixes

User Search over uploaded documents: Fixed an issue where the orchestrator failed to preserve uploaded file search context when using the Responses API, causing the assistant to incorrectly report that document content was "not found".

Language Model

New Model Deployments

Admin Mistral family of models now available for Tool Selection Backend now accepts union-type tool_choice (string and named function object). Named tool selection is automatically translated for Mistral models, validated across all providers (OpenAI, Azure, Mistral).

Admin GPT-5.4 is now available both through Azure and LiteLLM: For Azure OpenAI, use AZURE_GPT_54_2026_0305, for OpenAI via LiteLLM, use either litellm:openai-gpt-5-4 for GPT-5.4 or litellm:openai-gpt-5-4-pro for GPT-5.4 Pro.