ChatGPT’s New Updates Are Mind Blowing: ChatGPT Work, Plugins, Skills, Sites, and GPT 5.6

ChatGPT just got one of its biggest upgrades ever, and honestly, this changes a lot. We are not talking about one small feature release. We are talking about ChatGPT plugins, custom MCP connections, ChatGPT Skills, ChatGPT Work, ChatGPT Sites, a merged ChatGPT and Codex app, and new GPT 5.6 models all landing at once.

If you use AI for content creation, automation, research, coding, business workflows, or just getting more done faster, these updates are a major shift. ChatGPT is moving from being a chatbot you prompt into something much closer to a connected work system that can remember context, use tools, automate actions, and even build websites.

That is why this update feels so important. It is not just smarter answers. It is more capable execution.

Why this ChatGPT update matters

The big story here is that OpenAI rolled out a bunch of features that bring ChatGPT much closer to a true all in one productivity platform.

Here is what stands out most:

Plugins make it possible to connect ChatGPT directly to external tools.
Custom MCP support lets you hook ChatGPT into tools that are not natively listed.
Skills allow you to save repeatable workflows, writing styles, and SOPs.
ChatGPT Work turns ChatGPT into a task oriented workspace for getting things done.
ChatGPT Sites lets you generate live websites and web apps directly inside ChatGPT.
GPT 5.6 model options introduce new performance choices for speed, effort, and intelligence.
The desktop app merge with Codex brings work and developer workflows together.

Put simply, ChatGPT is becoming far more useful across real workflows, not just isolated prompts.

ChatGPT Plugins are the biggest upgrade yet

The first thing that jumps out is the new plugins experience. Inside ChatGPT, you can now see public plugins, personal plugins, and installed tools in a much more practical interface.

Featured plugins already cover a wide range of categories, including tools related to:

Data and analytics
Investment banking and public equity research
Gmail
Google Calendar
Figma
Slack
Outlook
Browser based actions

This matters because it moves ChatGPT from being a place where you think about work to a place where you can actually interact with the tools your work depends on.

Instead of constantly switching tabs, copying data, and pasting context into separate tools, ChatGPT can start operating with those tools directly connected.

That is the real shift.

Custom MCP access makes ChatGPT dramatically more flexible

What makes this even more powerful is the ability to add custom MCP access right from ChatGPT without needing to mess around with developer mode.

This means you are not limited to whatever appears in the plugin directory. If a tool has a custom MCP setup, you can connect it and use it inside ChatGPT.

A great example is connecting ChatGPT to vidIQ through a custom MCP. Once connected, ChatGPT can confirm that the integration is live and start using the tool right away.

One practical use case was pulling outliers in the cloud code niche. After the connection was in place, ChatGPT could surface structured results with filters like:

Title
Channel
Views
Views per hour
Newest
Outlier score

That is a big deal because the output is not just plain text. It becomes interactive and useful inline. Different plugins can bring their own interfaces and actions into the ChatGPT experience.

This is exactly the kind of thing that unlocks richer workflows for research, operations, marketing, and analysis.

ChatGPT Skills let you save repeatable workflows

The next major feature is ChatGPT Skills. This might end up being one of the most practical additions for people who do recurring work.

You can search skills, create skills, upload skills from your computer, build them through chat, or define them manually in an editor.

Here is where it gets exciting.

Imagine you upload multiple examples of how you structure your YouTube scripts. Then you ask ChatGPT to create a new script on a topic like how the YouTube algorithm works in 2026, using your voice and format.

ChatGPT can analyze the uploaded files, understand your structure, and generate a result in that style. Then you can tell it to turn that process into a YouTube script skill that you can call whenever you want.

From that point on, instead of reuploading examples and reexplaining your process, you can simply say something like:

Use my YouTube script skill for this topic

Now ChatGPT already knows the workflow, the format, the style, and the SOP behind it.

This is huge for:

Content creation
Email writing
Sales messaging
Research reports
Meeting summaries
Internal SOP generation
Client deliverables

Skills turn repeat prompting into reusable systems.

The new Skills editor

If you want more control, there is also a dedicated Skills editor. You can create a name, write a description, and define the actual instructions for the skill.

That gives you both options:

Generate a skill naturally through conversation
Build a skill precisely through an editor

For anyone building personal AI workflows, this is a very serious upgrade.

ChatGPT and Codex are now merged in the desktop app

Another important change is the new desktop app experience. ChatGPT and Codex have been brought together, which means the app now supports both general work and developer oriented use cases in one place.

In the top left area of the app, you can now switch between:

Work for getting work done
Codex for development workflows

This is more than a branding change. It reflects how OpenAI is organizing ChatGPT around actual use modes.

The interface itself also feels more unified and capable. Inside the app, plugins now include options like:

Computer use
Visualize
Browser access

There is also a much cleaner command driven experience. Typing a slash opens up quick controls for settings, personalization, projects, model switching, feedback, dictation, skills, and plugins.

That sounds small, but it removes friction. The faster you can control context, models, tools, and actions, the more useful ChatGPT becomes during real work.

GPT 5.6 is arriving with multiple model variants

One of the most talked about parts of this release is GPT 5.6.

The rollout appears to be staggered, with some people seeing it in the app before it shows up on desktop. The model options shown include:

5.6 Sol
5.6 Terra
5.6 Luna

There are also controls for effort and speed, which suggests a more explicit tradeoff system between fast responses and deeper reasoning. It is worth keeping an eye on usage because different combinations of model, effort, and speed appear to affect consumption differently.

What is clear already is that the newer models feel faster at carrying out complex tasks, especially when they involve reading context, using tools, and producing structured output.

Voice mode has a major overhaul too

The voice experience also got a substantial update.

Inside voice settings, you can now choose between different model types such as:

Live
Advanced
Standard

Under intelligence, there are additional choices like:

Instant
Medium
High

You can also select language and browse a wider set of voices. The overall result is a more transparent and customizable voice setup, especially with the new live model making the experience more powerful.

If voice is part of your workflow, this is not a minor refresh. It is a meaningful usability upgrade.

Zapier MCP connects ChatGPT to 9,000 plus apps

Even though plugins are powerful, there will still be tools you want that are not available natively. That is where Zapier MCP becomes incredibly useful.

By adding Zapier as a custom MCP connection inside ChatGPT, you can get access to more than 9,000 apps. That massively expands what ChatGPT can automate.

This helps solve one of the biggest limitations of any native plugin ecosystem. There will always be missing apps.

For example, direct plugin searches might not surface what you need for:

Instagram
YouTube
Apify
Square
School related tools

But inside Zapier, those connections can still exist, complete with their own triggers and actions.

That means ChatGPT can do more than answer questions about your systems. It can start helping define what should be automated based on the actions and triggers available across your apps.

If your goal is to build AI automations that stretch across your stack, this is one of the most practical additions in the whole update.

Cloud Browser gives ChatGPT browser level actions

Another feature worth exploring is Cloud Browser. This allows ChatGPT to open websites and take actions online, with permission controls built in.

You can manage site permissions and clear browser history when needed. That adds an important trust and control layer.

Once enabled, ChatGPT becomes much more capable in workflows that involve navigating the web instead of just summarizing it.

ChatGPT Work is designed for getting things done

If there is one feature that ties all of this together, it is ChatGPT Work.

This mode is clearly built around execution. It already has memory, can work with connected tools, and is framed as a workspace for real tasks rather than casual chatting.

ChatGPT Work can help with things like:

Researching and analyzing information
Working across connected apps and files
Creating documents, spreadsheets, presentations, reports, and sites
Following progress on multi step tasks
Answering questions during execution
Changing direction mid task
Approving important actions before they happen

There is also support for scheduled tasks, including tasks that repeat on a schedule, run from a trigger, or monitor for changes.

That makes ChatGPT Work feel much closer to an agentic productivity system.

ChatGPT Sites can build live websites and web apps

This is one of the coolest parts of the release. ChatGPT Sites lets you turn ideas into live websites directly inside ChatGPT.

You can simply describe what you want, grant the appropriate access, and let ChatGPT build it.

Possible use cases include:

Dashboards
Project trackers
Launch calendars
Prototypes
Internal portals
Reports
Simple interactive web apps

One example was creating a landing page for a TikTok growth and monetization accelerator. Another was creating a playable site experience. The important thing is not just that it generates code. It can produce something functional, previewable, and refined inside the workflow.

In one case, the generated project came together in just a few minutes and looked polished, functional, and surprisingly complete. That speed matters, especially for prototyping ideas that would otherwise sit on a to do list for weeks.

Browser import makes Work even more useful

There is also an option to import browser history, passwords, and cookies into ChatGPT Work. That means it can inherit more of your existing online context instead of needing to relearn every environment from scratch.

For the right workflows, that can make setup dramatically faster.

Small changes that make a big difference

Some of the most useful improvements are not flashy, but they matter a lot in day to day use.

Faster model behaviour when processing uploaded context
Better inline controls inside chats
Improved model switching
Cleaner personalization and project management
More direct access to skills and plugins
Mobile and desktop support for key new features

These are the kinds of improvements that make advanced features actually usable instead of just technically available.

My take: ChatGPT just became much more competitive

This release feels like OpenAI bundled months worth of platform improvements into one wave. Plugins, skills, workspaces, browser actions, app connections, website generation, and richer model controls all point in the same direction.

ChatGPT is becoming a serious operating layer for knowledge work.

That is why this feels bigger than a normal feature drop. The product is becoming more connected, more action oriented, and more reusable.

If you are only using ChatGPT as a place to ask one off questions, you are missing the real power of what it is becoming.

FAQ

What is ChatGPT Work?

ChatGPT Work is a task focused workspace inside ChatGPT designed to help with execution. It can research, analyze, create documents and sites, use connected apps, follow progress, and run scheduled tasks.

What are ChatGPT Skills?

ChatGPT Skills are reusable workflows or SOPs you can save inside ChatGPT. They let you repeat tasks such as writing in a certain style or following a consistent process without reexplaining everything each time.

What is custom MCP in ChatGPT?

Custom MCP lets you connect ChatGPT to external tools beyond the built in plugin library. It gives you a way to bring additional systems into ChatGPT directly, often without needing a separate developer setup.

What is ChatGPT Sites?

ChatGPT Sites is a feature that allows you to create live websites and web apps from prompts inside ChatGPT. It can generate, preview, and refine interactive pages without leaving the platform.

What are the GPT 5.6 models mentioned in the update?

The update shows GPT 5.6 model variants including 5.6 Sol, 5.6 Terra, and 5.6 Luna. These appear alongside controls for effort and speed, giving more flexibility in how tasks are handled.

How can ChatGPT connect to more than 9,000 apps?

By connecting Zapier MCP as a custom integration, ChatGPT can access a much wider app ecosystem. This expands automation options far beyond the native plugin list.

Meta description

Explore ChatGPT Work, Plugins, Skills, Sites, and GPT 5.6. See how these new ChatGPT updates unlock better automation, websites, and workflows.

Final thoughts

If you want to get the most out of these updates, start with three things: connect the tools you already use, build one or two repeatable skills, and test ChatGPT Work on a real project instead of a toy prompt.

That is where the difference shows up fast.

If you have been waiting for ChatGPT to feel less like a demo and more like a serious work platform, this is one of the clearest signs yet that it is heading there. Try the new features, experiment with connected workflows, and see which part of your process you can remove, speed up, or fully automate.

If this helped you, share it with someone exploring AI workflows, and check out the related guides above for more ways to use ChatGPT in real business and creative systems.