Changelog

ChatGPT 5.6 Luna· Jul 13

Replaces ChatGPT 5.5 Instant as the fast general-purpose OpenAI model, with lower cost and strong everyday chat, tool use, and workflow performance on Azure.

ChatGPT 5.6 Terra· Jul 13

Replaces ChatGPT 5.5 Thinking as the reasoning OpenAI model, with stronger multi-step reasoning, tool use, and long-context performance on Azure. Also available in Agent Builder under Sol.

ChatGPT 5.6 Sol· Jul 13

Premium reasoning model in the 5.6 family for the hardest multi-step tasks, deep tool use, and complex professional workflows on Azure. Replaces ChatGPT 5.5 in Agent Builder.

Chat sharing· May 11

Conversations from your private assistant can now be shared, either publicly or with selected teammates from your organization. Choose between view-only access (recipients can read the conversation) or full access (recipients can continue the conversation, ask follow-up questions, and re-share with others). Just click the Share icon at the top right of any chat to get started.

Image & video sharing· May 11

Share chats containing generated images or videos just like any other conversation, publicly or with specific members of your organization. With view-only access, recipients can see the result, with full access, they can regenerate or edit the image or video, and re-share with others.

Delete image· May 5

You can now delete an image from a conversation or the gallery without deleting the entire conversation.

Grok 4.3· May 7

Upgrades Grok 4.2 with stronger reasoning, better instruction following, improved accuracy and tool calling, plus a large 1M-token context window for complex coding, research and agentic workflows.

Qwen 3.6 Flash· May 4

Faster, more cost-effective version of Qwen 3.6, offering near-flagship performance with the same long-context and multimodal capabilities, optimized for production use at lower cost.

Qwen 3.6 Plus· May 4

Alibaba's flagship Qwen 3.6 model, built for agentic coding, complex reasoning, multimodal understanding, and long-context workflows with a 1M-token context window.

DeepSeek V4· May 4

Major upgrade from V3.2, adding a more efficient hybrid attention architecture, 1M-token context, open-weight Pro and Flash versions, and stronger long-context, coding, reasoning, and cost-performance capabilities.

Seedance 2.0· May 1

Now available in Unitalk. ByteDance's next-generation video generation model, supporting text, image, audio, and video inputs to create cinematic videos with native audio, realistic motion, and multi-shot editing.

CSVs & Excel analysis· May 1

Fixed an issue in file analysis when consolidating multiple files.

Desktop app download· April 21

Unitalk is now available as a downloadable desktop app. As a Progressive Web App, you can install it on your computer for quick access, a native-like experience, and seamless use across your devices.

Extend Video mode (Veo 3.1 & Veo 3.1 Fast)· April 17

Extend your videos by up to 8 additional seconds using Google’s Veo 3.1 and Veo 3.1 Fast models. Simply describe what happens next, and your video is seamlessly extended — perfect for creating longer sequences with consistent characters. To use it, click the "Extend Video" icon below any video generated with Veo 3.1 or Veo 3.1 Fast.

Introducing AI Agents: Iris, Lumi, Voxo & Otto· April 6

The "Create" menu has evolved into a team of specialized AI agents. What used to be separate Image, Video, Audio, and Agents menus are now dedicated agents — each with a name, a role, and a clear expertise: Iris (Graphic designer), Lumi (Videographer), Voxo (Voice Assistant), and Otto (Automation Specialist). This is the first step toward a full team of autonomous AI agents working alongside you. More agents coming soon.

One-click agent templates· April 3

Install ready-to-use automation templates in one click from the Agent menu.

Revamped image menu· April 6

New menu with a template gallery organized by category and a clear distinction between text-to-image and image-to-image.

Revamped video menu· April 6

Same revamp for video: templates organized by category with clear text-to-video and image-to-video modes.

ChatGPT 5.5· April 28

Improves on GPT-5.4 with stronger reasoning, coding, research, tool use, and long-context performance for complex professional workflows.

ChatGPT Images 2.0· April 27

OpenAI's latest image model, with higher-quality visuals, better text rendering, improved editing, and stronger realism/style control.

Claude Opus 4.7· April 22

Latest model from Anthropic, notably improved over Opus 4.6 for advanced software engineering, with especially strong gains on the most difficult tasks.

Veo 3.1 Lite· April 13

Google's most cost-effective Veo model, built for high-volume video generation with fast speed, lower cost, and support for 720p/1080p output, but no 4K.

Kimi 2.5 (Moonshot)· April 13

Now available in the Automation Builder, designed for advanced agentic workflows.

Minimax 2.7· April 13

A powerful new model optimized for agent-based automation, now available in the Automation Builder.

Veo 3.1 Fast· April 2

A faster, more cost-effective version of Veo 3.1, ideal for quicker iterations while maintaining strong video quality.

Kling 3.0· April 1

The Kling video generation model has been upgraded to version 3.0 for higher quality results.

Regenerate images· April 30

Fixed issues affecting image regeneration.

Affiliate Program· March 30

Earn 30% recurring commission on every customer you refer to Unitalk. Join our affiliate program and start earning today.

Unitalk Google Sheets & Microsoft Todo, Outlook, OneDrive, Excel 365, Sharepoint nodes· March 17

Available in n8n with one-click authentication via Unitalk.

Favorite conversations· March 12

Mark your conversations as favorites to find them instantly.

AI credits usage export· March 13

Download a CSV summary of your AI credit usage from your Billing section.

Unitalk Google Calendar & Google Drive nodes· March 13

Available in n8n with one-click authentication via Unitalk.

Unitalk Gmail node· March 12

Connect your Gmail in one click via your Unitalk account directly in your n8n workflows.

Complex PDF support (OCR)· March 18

PDFs containing images or scans are now better supported thanks to optical character recognition (OCR).

Model updates· March

ChatGPT 5.4 Thinking, ChatGPT 5.3 Instant, Nano Banana 2, Gemini 3.1 Flash — all available now.

YouTube transcript retrieval fix· March 23

File analysis in custom assistants fix· March 18

App connection from the Agent Builder fix· March 17

Deep Research· February 27

Launch in-depth research on complex topics and get detailed, sourced answers with the new Deep Research mode.

Agent Builder — Canvas view· February 24

Build your automations visually with the new Canvas view in the Agent Builder.

Audio recording· February 20

Record your voice directly from the Audio menu to dictate notes or instructions.

Extended file format support· February 6

Your knowledge base now accepts a much wider range of file formats. Full list of supported formats :

Prompt Enhancer· January 30

Automatically improve your prompts: the Prompt Enhancer in new chat menu rewrites and enriches your instructions for better results.

Custom instructions — Improved saving· February 26

Editing and saving your custom instructions is now more reliable.

Simplified app connection in the Agent Builder· February 5

Connect your apps in one click via Unitalk when building your workflows — no manual setup in n8n required.

Unsupported format error message· February 6

A clear message now lets you know when you upload a file format which is not supported.

Model upgrades· February

Gemini 3.1 Pro, Claude Opus 4.6, Veo 3.1 — upgraded from Gemini 3 Pro, Claude Opus 4.5 and Veo 3

Various stability fixes· February

Google Calendar MCP, custom instructions, CSV/Excel file search and analysis.

Agent Builder — Prompt-based automation· January 7

Describe the automation you want in plain language: the Agent Builder builds and deploys it automatically on your n8n instance.

Self-hosted n8n instance connection· January 14

Connect your own self-hosted n8n instance to your Unitalk account to use your existing infrastructure.

Custom assistant settings redesigned· January 12

The configuration interface for your custom assistants has been redesigned for a simpler setup experience.

Model upgrades· January

Veo 3.1 upgraded from Veo 3.

One-click private server· December 23

Deploy a private server with n8n pre-installed in one click from your account.

New interface· December 22

The assistant interface has been completely refreshed with a new, modern design.

Model upgrades· December

Deepseek V3.2, GPT-5.2, ChatGPT Image 1.5, Mistral Large 3 - upgraded from Deepseek V3.1, GPT-5.1, ChatGPT Image.