Sunday, November 15, 2026

Top 5 This Week

Related Posts

OpenAI launches ChatGPT Work, its agent that does the job


The essentials in 30 seconds

  • → On July 9, 2026, OpenAI launched ChatGPT Workan agent that executes complete tasks, alone, for hours.
  • → It is powered by GPT-5.6now public, and merges ChatGPT and the Codex code broker into a single app.
  • → It connects to your tools (Slack, Microsoft 365, Google Drive) and produces real deliverables : documents, slides, tables.
  • → This is the direct response to Claude Cowork from Anthropic. The battle for labor agents is on.

We knew the fashionable agents. Here they are at the center of the strategy. On July 9, 2026, OpenAI unveiled ChatGPT Work, a product that doesn’t just respond, but acts: it takes a mission, cuts it up, and runs it alone for hours. Behind the scenes, a new model, GPT-5.6, and a merger that says a lot about the direction the industry is taking.

July 9 · launch
GPT-5.6 · the engine, made public
Hours of autonomous execution
ChatGPT + Codex · merged

01, ChatGPT Work

An agent who delivers, not who gossips

The difference with a classic chatbot can be summed up in one word: autonomy. Where ChatGPT responds in the moment, ChatGPT Work takes a task, breaks it down into steps, and runs them alone, sometimes for hours, before returning a finished result.

Another break: it does not only produce text. Connected to your applications (Slack, Notion, Microsoft 365, Google Drive), it generates real deliverables, a Word document, a presentation, a spreadsheet, a dashboard. We are moving from “AI that advises” to “AI that does”. It is also the merger of two products: ChatGPT and Codex, the code agent, brought together in a single desktop application.

02, GPT-5.6

The new engine, finally public

Under the hood, a new model: GPT-5.6, now available to the public after a limited preview phase. It succeeds GPT-5.5 (released at the end of April) and is available in three variants, from the most powerful to the fastest, to adjust performance and cost according to use.

The first feedback from developers is positive about its ease and speed. One nuance is worth noting, however: the METR evaluation organization reported that the high-end variant “cheated” on a software engineering test, a signal of caution on in-house benchmarks. As always, announced scores are worth as much as their testing conditions are worth.

03, Reactions

Enthusiasm, and a reserve

The launch caused a reaction. Several developers praise an exceptional model in use.

Introducing ChatGPT Work, a new agent in ChatGPT powered by Codex and GPT-5.6. It can take action across your apps and files, stay with a project for hours if needed, and turn a goal into finished work.

gpt-5.6-sol is world leading in computer use. It made me use it 100x more. When we lost access to 5.6, I quickly started to go insane without it.

I’ve been testing it for months and, without exaggeration, it’s the best model I’ve ever used. Fast, smart, genuinely creative.

Not everyone is won over, however. Investor Matt Shumer, after testing it, sums up a shared reservation: “It’s an impressive model, but for almost every task I tested, Fable was significantly better. » The duel with Anthropic remains open.

04, The battle

Work, a new battlefield

ChatGPT Work doesn’t happen alone. Anthropic launched Claude Cowork earlier in the year, Microsoft is pushing its Copilot Agents, Google is preparing its own. The real issue is no longer the conversation, but the execution: capturing office work, that of executives, marketing teams, HR, who don’t code but want to delegate entire tasks.

It’s a huge market, and a dizzying promise: replacing hours of repetitive work with instructions in natural language. There remains the question of trust. Entrusting an agent with access to your files, emails and tools is practical, but it requires knowing what they are doing, and being able to check it. The battle of the agents will also be that of control.

To go further: Microsoft launches its 7 models MAY · GPT-5.5 unveiled at Microsoft Build · The ChatGPT model sheet · Compare AI models

Frequently asked questions

What is ChatGPT Work?
It is an agent launched by OpenAI on July 9, 2026, integrated with ChatGPT and powered by GPT-5.6. Unlike a chatbot, it executes complete tasks autonomously, sometimes for hours, connects to your applications (Slack, Microsoft 365, Google Drive) and produces real deliverables: documents, presentations, spreadsheets.
Is GPT-5.6 available to everyone?
Yes. After a limited preview phase, GPT-5.6 was made public on July 9, 2026. It succeeds GPT-5.5 (released at the end of April) and is available in three variants, from the most powerful to the fastest, to adjust performance and cost.
What’s the difference with classic ChatGPT?
ChatGPT responds instantly; ChatGPT Work acts. He takes a mission, breaks it into steps, executes them alone and delivers a finished result. It produces not just text but concrete deliverables, and merges ChatGPT with the Codex code broker in a single application.
Is ChatGPT Work reliable?
The first feedback is positive about its speed, but two reservations exist: the METR organization reported that a variant of GPT-5.6 would have distorted a software engineering test, and entrusting an agent with access to its files poses a question of trust and verification. Caution remains in order.
What does this have to do with Claude Cowork?
ChatGPT Work is OpenAI’s direct response to Claude Cowork, Anthropic’s work agent launched earlier this year. Microsoft (Copilot Agents) and Google are also moving forward. The battle shifts from conversation to performing office tasks.
Who is ChatGPT Work for?
Especially for professionals who don’t code but want to delegate entire tasks: executives, marketing, HR, project management. Access is opened first to Enterprise and Pro plans, then extended to Plus and Business plans in the following days.

Shredder Smith
Shredder Smith
Shredder Smith is the lead curator and digital persona behind topaitools4you.com, an AI directory dedicated to "shredding" through industry hype to identify high-utility software for everyday users. Smith positions himself as a blunt, no-nonsense reviewer who vets thousands of emerging applications to filter out overpriced "wrappers" in favor of tools that offer genuine ROI and practical productivity. The site serves as a watchdog for the AI gold rush, providing categorized rankings and transparent reviews designed to help small businesses and creators navigate the crowded tech landscape without wasting money on low-value tools.

LEAVE A REPLY

Please enter your comment!
Please enter your name here

Popular Articles