On July 9, 2026, OpenAI announced ChatGPT Work, a new autonomous AI agent inside ChatGPT, and began rolling it out globally. Built on Codex technology and the GPT-5.6 model, it takes action across apps and files and can stay with a project for hours when needed.
July 9, 2026 · OpenAI
ChatGPT Work: the Agent That Finishes the Job
A new autonomous agent inside ChatGPT — built on Codex and the GPT-5.6 model — takes action across apps and files, staying with a project for hours to deliver finished documents, spreadsheets, decks and Sites.
5M+
weekly Codex users powering the shift
1M+
of them in non-developer work
3
unified modes: Chat · Work · Codex
24h
to planned full global deployment
Sol vs Claude Fable 5 — Agents' Last Exam
The GPT-5.6 flagship claims a +13.1 point lead on reasoning.
Sol also scores 80 on the Coding Agent Index (+2.8), and can run sub-agents in parallel under Ultra.
GPT-5.6 tiers
Sol
Flagship · reasoning-focused. State-of-the-art agentic + coding results.
Terra
Balanced tier for everyday work.
Goal → finished deliverable
Give a goal
→
Act across apps & files (hours if needed)
→
Docs, sheets, decks, Sites
Early results in the field
40%
less time on GTC prep — NVIDIA
weeks → hours
competitive analysis — Virgin Atlantic
Automated lead review & release checks
Zapier · RingCentral
Praise
Codex + ChatGPT unify workflows in one app
Sol's persistence and stronger intent understanding
Concerns
Regular Chat demoted to a Quick chat button
Cloud Work and desktop Work don't sync at launch
Mixed feelings on differences from 5.5
Staged rollout
Web / mobile Work
Pro / Enterprise / Edu first; Plus / Business within days
Continue reading The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.
Already purchased? Sign in ✓ Signed in — this article isn’t included in your current plan.Unlocking the full article…