GILD verdict
Choose Claude where writing is the main job, and ChatGPT where people already rely on it and its memory.
- The workflow
- Choosing a company assistant for writing, research and agent work across a team of 2 to 200
- The decisive evidence
- Reported4 of 5 named teams run both ChatGPT and Claude; What teams report below sets out each account.
- ReportedTiny's co-founder uses Claude for writing, which he rates best, Gemini for very large inputs and ChatGPT for everything else.16
- ReportedGiven the same Gmail task, ChatGPT Work sent an email that Claude Cowork only drafted, because of permissions Ethan Mollick had set earlier.13
- DocumentedStandard and Premium seats cost the same on both, and at 200 seats Claude Enterprise comes to US$4,560 a month against US$4,000 for ChatGPT Business by GILD's calculation from published rates.18
- GILD analysisNone of these accounts measured output on the same tasks, so they show preference, and a measured pilot still decides.
- The tradeoff
- Run both with one team for a month before you standardize. Claude's Enterprise bill rises with use above 150 seats, while ChatGPT Enterprise is priced through sales with credit or token billing set by the agreement.
- What would change the call
- If your pilot team keeps opening the other assistant, license that one for the team instead.
Named teams keep both: Brex buys a few seats of each and lets staff choose, and Every's CEO opens ChatGPT first for its memory, while Tiny's co-founder writes in Claude. Seat prices match at US$20 billed annually for Standard and US$100 for Premium; above 150 seats Claude Enterprise adds usage at API rates, US$4,560 against US$4,000 a month for 200 seats by GILD's calculation.
What teams report
Run both 4
Brex
Brex chose not to standardize, so it buys a few seats of ChatGPT, Claude and Gemini and lets each employee request one through ConductorOne in Slack (January 17, 2026).
“And so, for instance, like we allow employees to... basically go to, in Slack and use Conductor One to get a ChatGPT, a Cloud or a Gemini license.”
James Reggio, CTO, BrexLatent Space, , around 21:15 Every
Shipper reports that o3 in ChatGPT is the first thing he opens, and that memory is what still sets it apart from Claude for him (July 17, 2025).
“And what it has that really makes a difference still from Claude is it has memory.”
Dan Shipper, Co-founder and CEO, EveryLenny's Podcast, , at 36:03 Zevi Arnovitz, Product manager
Arnovitz keeps projects in both; he says ChatGPT's memory mixed unrelated contexts, and when Meta approached him about a role he prepared in a new Claude project (January 18, 2026).
“And what happened was the memory feature was mixing stuff up.”
Zevi Arnovitz, Product manager, MetaLenny's Podcast, , around 8:34 Tiny
Wilkinson uses Claude for writing, which he rates best, Gemini for large inputs and ChatGPT for everything else (July 3, 2025).
“Claude, I use for writing very extensively. I find it's the best for that.”
Andrew Wilkinson, Co-founder, TinyLenny's Podcast, , at 51:25
Evaluated 1
Ethan Mollick, Professor
Given the same seminar-preparation task over Gmail in July 2026, ChatGPT Work sent an email to his colleagues while Claude Cowork drafted one, because of permissions he had set earlier.
“But you may have noticed something; Claude (the top response) only prepared a draft but ChatGPT actually sent an email to my colleagues! What happened? Well, it was my fault. I had previously given ChatGPT permission to send email on my behalf, and Claude was told to ask me first.”
Ethan Mollick, Professor, Wharton SchoolOne Useful Thing, , write-up
Choosing between ChatGPT and Claude
Choose ChatGPT if
- You want ChatGPT Work, record mode, and Sites in the same workspace as chat, with Word, Excel, PowerPoint, and Google Sheets extensions.
- You need data residency choices across ten regions, listed for Enterprise.
- Your people already use ChatGPT, and you want their work inside a managed workspace.
- People rely on memory across chats, which Every's CEO names as what keeps him opening ChatGPT first.
Choose Claude if
- You want Enterprise priced as US$20 per seat per month billed annually plus usage at API rates.
- Your teams work in Microsoft 365 and want Claude for Microsoft 365, Outlook, and Chrome, which the Team plan lists.
- Teams of 2 to 150 want a self-serve plan with a path to self-serve Enterprise.
- Writing is the main job, which Tiny's co-founder gives Claude because he rates it best for writing.
Side by side
| Dimension | ChatGPT | Claude |
|---|---|---|
| Individual plansList prices in US dollars. | Free. Go US$8. Plus US$20. Pro US$100, US$200, or US$500 per month, billed monthly | Free. Pro US$20 per month, or US$17 per month billed annually as US$200 up front. Max 5x US$100 and Max 20x US$200 per month |
| Team seats | Business Standard US$20 per user per month billed annually or US$25 monthly. Premium US$100 or US$125. Two seats minimum | Team Standard US$20 per seat per month billed annually or US$25 monthly. Premium US$100 or US$125. Teams of 2 to 150 |
| Enterprise pricing | Custom pricing through sales, with credit-based or token-based billing depending on the agreement | US$20 per seat per month billed annually, plus usage at API rates. Self-serve or sales-assisted |
| Training on personal plans | New conversations can be used for training unless Improve the model for everyone is turned off. Temporary chats are excluded | Listed as opt-out on Free, Pro, and Max. Chats are used for training only when the model improvement setting allows it, and incognito chats are excluded |
| Training on business plans | Business, Enterprise, and Edu content is not used for training by default | No model training on Team and Enterprise content by default |
| Retention | Enterprise admins set retention. Business admins control retention. Deleted conversations leave OpenAI systems within 30 days | Custom data retention controls on Enterprise. Deleted conversations leave Anthropic's back end within 30 days |
| Admin and compliance | Business: SAML SSO, MFA, admin console. Enterprise adds SCIM, role-based access, EKM, IP allowlisting, and the Compliance API | Team: SSO and central billing. Enterprise adds SCIM, role-based access, audit logs, the Compliance API, IP allowlisting, and a HIPAA-ready offering |
| Agents and connectors | ChatGPT Work, scheduled tasks, workspace agents, plugins and apps, record mode on macOS, and Dots on Pro and Business Premium | Connectors, skills, hand-off and scheduled tasks, Claude in Chrome, and enterprise search on Team and Enterprise |
| 200 seats a monthSeats only; no purchased credits, taxes or negotiated discounts. Claude's usage share grows with real use. | US$4,000 on Business Standard at US$20 a seat billed annually. GILD's calculation from published rates, read Oct 28, 20261 | US$4,560 on Enterprise, above Team's 150-seat cap: US$20 a seat plus 200 requests a seat of 2,000 input and 1,000 output tokens on Sonnet 5.5 at US$2 and US$10 per million. GILD's calculation from published rates, read Oct 28, 20268 |
| What named teams did | Every's CEO opens ChatGPT first for its memory; Brex buys some seats of each and lets staff choose1214 | Tiny's co-founder uses Claude for writing, which he rates best; a product manager prepared for a Meta role in a Claude project1516 |
What practitioners report, side by side
Each product's figures come from its own practitioners, for the group you pick.
Overall signal
Leaders and teams rate Claude +44, 30 points above ChatGPT at +14. Most of the gap is in complaints, with 10 of ChatGPT's 38 sources negative against 2 of Claude's 28.
The score counts the authors from leaders and teams that take a side, one vote each with their latest view, and is how far positive votes lead negative ones, per 100 votes, so ChatGPT's +14 rests on votes split 15 positive, 11 mixed and 10 negative and Claude's +44 on 14, 11 and 2.
A lead of a fifth of the votes reads positive and a lead of half strong positive, with the same lines below zero. The dots show each source, positive, mixed, negative. Run a pilot on your team's recurring tasks to measure how often answers are used without rework.
Footprint on the AI Knowledge Work stack
They differ on one of 7 layers. On Meeting notes, ChatGPT supports it and Claude touches it. Coverage is each product's placement on this stack, as its product page shows it. Core is a job the product owns; Supports and Touches are jobs it helps with. Coverage says nothing about how well a product does them.
| Layer | ChatGPT | Claude |
|---|---|---|
| Create | ||
| Meeting notes | Supports | Touches |
The 6 layers where they match
| Layer | ChatGPT | Claude |
|---|---|---|
| Assist | ||
| AI assistant | Core | Core |
| Find | ||
| Web research | Supports | Supports |
| Enterprise search | Supports | Supports |
| Create | ||
| Documents | Supports | Supports |
| Analyze | ||
| Data analysis | Supports | Supports |
| Automate | ||
| Automation | Supports | Supports |
What practitioners praise
Among leaders and teams, Hand-off Tasks and Connectors & Integrations reach 5 sources from 3 authors only for Claude, at 7 of 7 and 5 of 8. ChatGPT needs a pilot on the same tasks before you choose on either. Custom GPTs, Projects & Memory reaches 5 sources from 3 authors only for ChatGPT, at 3 of 5. Claude needs a pilot on the same tasks before you choose on it.
| Praised for | ChatGPT | Claude |
|---|---|---|
| Custom GPTs, Projects & Memory | 3 of 5 | Not a theme in its sources |
| Hand-off Tasks | Not a theme in its sources | 7 of 7 |
| Connectors & Integrations | Not a theme in its sources | 5 of 8 |
Each count is the sources that praise a theme, or raise it as a concern, out of those that discuss it. A row pairs the two products when both carry a theme under the same or an equivalent name.
A count ending in "so far" has under 5 sources from 3 authors among leaders and teams. "Not a theme in its sources" means GILD found no theme of that name in that product's sources.
What practitioners raise as concerns
Among leaders and teams, Workspace Administration and Data Privacy & Retention reach 5 sources from 3 authors only for ChatGPT, at 7 of 10 and 3 of 8. Claude needs a pilot on the same tasks before you choose on either. Both carry Usage Limits & Cost. It is raised in 6 of the 9 ChatGPT sources that discuss it, against 5 of 7 for Claude. Missing Features and Data & Admin Controls reach 5 sources from 3 authors only for Claude, at 6 of 6 and 2 of 7. ChatGPT needs a pilot on the same tasks before you choose on either.
| Raised as a concern | ChatGPT | Claude |
|---|---|---|
| Usage Limits & Cost | 6 of 9 | 5 of 7 |
| Workspace Administration | 7 of 10 | Not a theme in its sources |
| Data Privacy & Retention | 3 of 8 | Not a theme in its sources |
| Missing Features | Not a theme in its sources | 6 of 6 |
| Data & Admin Controls | Not a theme in its sources | 2 of 7 |
16 sources
- ChatGPT pricing
- OpenAI Help Center: ChatGPT Business overview
- OpenAI Help Center: About ChatGPT Pro tiers
- OpenAI Help Center: What is ChatGPT Enterprise?
- OpenAI: Enterprise privacy at OpenAI
- OpenAI Help Center: Data controls in ChatGPT
- OpenAI Help Center: ChatGPT Record
- Claude pricing
- Anthropic Privacy Center: Is my data used for model training?
- Anthropic Privacy Center: How long do you store my data?
- Anthropic Privacy Center: How long do you store my organization's data?
- Latent Space: Brex's AI strategy, with James Reggio (Jan 17, 2026)
- Ethan Mollick: An opinionated guide to which AI to use (Jul 23, 2026)
- Lenny's Podcast: Dan Shipper on how Every works with AI (Jul 17, 2025)
- Lenny's Podcast: Zevi Arnovitz, the non-technical PM's guide to building (Jan 18, 2026)
- Lenny's Podcast: Andrew Wilkinson of Tiny (Jul 3, 2025)