이미지: METAL LAB 생성
Summary
- OpenAI's head of core products, Thibault Sottiaux, told TechCrunch in an interview that ChatGPT Work has surpassed 20 million users
- ChatGPT Work is included in the $20/month Plus plan, and Sottiaux said pricing has dropped 80% since the Luna optimization rolled out
- Sottiaux said the goal is to keep the interface minimal and let the model make its own judgment calls, a contrast he drew with Claude Cowork, which presents users with choices at every step
- 인터뷰이
- 티보 손티오, 오픈AI 핵심제품 총괄(API·에이전트 인프라·엔터프라이즈·챗GPT·코덱스)
- 보고 라인
- 그렉 브록만
- 챗GPT 워크 이용자
- 2000만명 (2026년 8월 발표 시점)
- 요금
- 플러스 요금제 월 20달러에 포함
- 가격 인하
- 루나 적용 이후 80% 인하, 영구적 조정이라고 설명
- 핵심 모델
- GPT-5.6 — 문서 처리·슬라이드·보고서·딥리서치 능력 향상
- 경쟁 비교
- 에단 몰릭(와튼스쿨)이 챗GPT 워크는 마법 지향, 클로드 코워크는 선택지 제공 방식이라 비교
- 발행
- 테크크런치, 2026년 8월 25일 인터뷰
Thibault Sottiaux, OpenAI's head of core products, says ChatGPT Work has crossed 20 million users. He shared the figure in an interview with TechCrunch — notable partly because Sottiaux also leads the coding tool Codex and reports directly to Greg Brockman. The conversation ranged widely: why ChatGPT Work matters, how it's priced, and the underlying product philosophy of building agents that do work on people's behalf.
From Codex to ChatGPT Work: the next phase of rollout
Among Codex users, Sottiaux is known as the guy who suddenly lifts token limits. Codex first launched for technically forgiving early adopters, and Sottiaux says OpenAI is now working to carry that same capability over to non-developers. He described this as a process of getting the technology to reach the widest possible population — which is why ChatGPT Work is bundled into the $20/month Plus plan rather than sold separately.
Minimal screen, maximum autonomy
The product philosophy Sottiaux emphasized is one of getting out of the model's way and letting it express itself. He said he'd rather resolve things through a single conversation than clutter the screen with buttons, pointing to ChatGPT Voice as an example — growth picked up noticeably, he said, once voice conversation was added to what had been a text-only ChatGPT.
That stands in sharp contrast to a competing product. Wharton professor Ethan Mollick, in a post on Bluesky, compared the two: ChatGPT Work aims for a magic-like experience, while Anthropic's Claude Cowork puts choices in front of users at every step, almost like an A/B test. Sottiaux's response to that comparison was, "The world seems to be ready" — and he pointed to the 20-million-user figure as evidence.
Pricing keeps moving
Sottiaux said OpenAI is pushing for efficiency gains every single day, and that pricing has fallen 80% since the Luna optimization was applied. He framed this not as a one-off discount but as an ongoing pricing adjustment, and set a goal that users should be able to do more with the same spend six months from now than they can today.
| Item | Detail |
|---|---|
| Users | 20 million (as of announcement) |
| Pricing | Plus plan, $20/month |
| Price cut | 80% since Luna |
| Core model | GPT-5.6 |
GPT-5.6 and product as discovery
Sottiaux said that every jump in model performance triggers a fresh round of discovering what's now possible. With GPT-5.6, capabilities that feel much closer to real work — processing massive documents, generating polished slides and reports, deep-research-style tasks — moved up a level, and OpenAI builds its products around that new ceiling. He called this part of an iterative deployment process: gathering community feedback and continuously refining from there.
Why open up email and iMessage
On the day of the interview, OpenAI shipped a new feature giving ChatGPT access to iMessage. Asked about concerns over this kind of expanding access, Sottiaux said the key is choosing a model you can trust to be safe. He pointed to OpenAI's investment in its safety stack and its track record of publishing honest benchmarks as evidence that its models lead the field on that front.
Editor's take
The most telling moment in this interview is that Sottiaux didn't dodge the Claude Cowork comparison. Anthropic has chosen to build trust by showing users their options; OpenAI has gone the opposite direction, minimizing intervention to make the experience feel like magic. Both companies are solving the same problem — getting non-developers to trust agents with real work — from opposite directions, and which UX philosophy wins is shaping up to be the defining experiment in the office-agent market going forward.
There's another interesting angle if you compare generations of the product. Codex first rolled out to technically forgiving developers, and that same capability has now carried directly over into ordinary office work like writing documents or building slide decks. Where automation once mainly benefited people who could write code, GPT-5.6 has put the same kind of automation into the hands of people who write reports for a living.
From a practical standpoint, two things are worth watching. One is pricing: if the ongoing adjustments Sottiaux described actually continue, teams currently worried about ChatGPT Work's token consumption could well be doing a lot more with the same budget within six months. The other is data access — as the push to open up email and iMessage continues, organizations would do well to set internal rules now about which accounts get connected to agents.
In the coming weeks, it wouldn't be surprising to see Claude Cowork respond to OpenAI's 20-million-user announcement with its own adoption numbers. The fact that Sottiaux addressed the competitor comparison head-on in this interview is itself a sign that OpenAI is well aware this is turning into a numbers fight.
Comments