ChatGPT Plus and Claude Pro both cost $20 a month, and both now come with a coding agent: Codex on Plus, Claude Code on Pro. Buy ChatGPT Plus for image generation, voice with video and chat that OpenAI lists as unlimited. Buy Claude Pro to work inside Word, Excel, PowerPoint and Outlook, read very long documents, and connect your own tools so Claude can act on them.
We read both vendors’ pricing pages and help centers on November 12, 2026, and put every line for the two $20 plans side by side: models, how the limits are worded, the coding agent and its limits, Office add-ins, files, projects, connectors, memory, images and voice. That table is the first section. The second is a chart of each vendor’s price ladder, from free to its top plan.
Plenty of people stop choosing and pay for both. We make Context Board, one workspace where Claude Code and Codex both work, with a widget that shows what is left of both plans side by side. Context Board is ours, so weigh the bias. The last section covers it.
- Same price: $20 a month each. Claude Pro also sells a year for $200 ($17 a month). ChatGPT Plus has no yearly option.
- Our table, from both vendors’ pages on November 12, 2026: Plus leads on images, voice with video, model choice and chat limits. Pro leads on Office add-ins, connectors that can make changes, context window and the yearly price.
- Limits: Claude Pro has a five-hour and a weekly limit, shared by chat and Claude Code. ChatGPT lists text chats as unlimited and meters Codex separately: roughly 15 to 160 local messages per five hours on GPT-6.1 Sol.
- Coding: Claude Code runs Opus 5.5 by default on Pro. Codex on Plus runs GPT-6.1 Sol and GPT-6 Luna, with cloud tasks for eligible accounts.
- Paying for both? Context Board shows both plans’ limits side by side, and Claude Code and Codex work on the same boards. Free for 3 days, then $12 a month.
What $20 a month gets you: ChatGPT Plus vs Claude Pro
| Claude Pro | ChatGPT Plus | |
|---|---|---|
| Price | $20 a month, or $17 a month billed yearly ($200 up front) | $20 a month, monthly only |
| Models | Opus, Sonnet and Haiku. Fable paid with usage credits | GPT-6.1 Sol, GPT-6 Astra, GPT-6 Sol, GPT-6 Luna, plus older models |
| Chat limit, in their words | ”More usage than Free, every 5 hours and every week” | Everyday text chats “Unlimited”, subject to abuse guardrails. “Expanded messages and uploads” |
| Coding agent | Claude Code, Opus 5.5 by default | Codex, GPT-6.1 Sol and GPT-6 Luna |
| Coding agent’s limit | Same five-hour and weekly pool as chat. No message count published | Own five-hour window, about 15 to 160 local messages on GPT-6.1 Sol. Weekly limits may apply |
| Office add-ins | Excel, PowerPoint, Word and Outlook | Excel, Word, PowerPoint and Google Sheets |
| Makes files | .xlsx, .pptx, .docx and PDF | Word, Excel, PowerPoint and Google files, in ChatGPT Work |
| Projects | Yes (Free gets up to 5) | Yes, plus scheduled tasks and custom GPTs |
| Your own MCP servers | Custom connectors that can read and make changes | Developer mode (beta). Full read and write MCP is for Business and Enterprise/Edu |
| Memory | Memory across conversations | Expanded memory, and memory with past chats |
| Images | None listed. Charts as PNG files | Image generation, also with Thinking |
| Voice | Voice mode | Expanded voice, and voice with video |
| Context window | Up to 1M, varies by model | 54K for GPT Instant, 256K for reasoning |
| Research | Research | Expanded deep research |
| Browser | Claude in Chrome | Built-in browser |
How we made it: we read Claude’s pricing page and ChatGPT’s pricing page on November 12, 2026, including each one’s full comparison table, then the help articles linked in the sections below. Words in quotes are the vendors’ own. Where a pricing page showed a local currency, we took the US price from the same vendor’s docs or help center.
Read across the rows and the two plans split cleanly. ChatGPT Plus is the broader consumer plan: pictures, voice with video, a long model list and chat with no posted cap. Claude Pro is the work plan: Office add-ins that edit your files, connectors that act on your tools, and a context window nearly four times the largest one Plus lists.
The price ladder, from free to the top plan
Both $20 plans sit on a ladder. The rungs above them matter, because the $20 plan is where you find out how much you use.
We counted the individual plans on each vendor’s own pages. Claude has four rungs, from Free to Ultra at $200. ChatGPT has six, from Free to Pro 500 at $500, with Go at $8 as a step below Plus that Claude doesn’t offer.
Claude, per its pricing page: Max at $100 gives “5x Pro usage” every five hours and every week. Ultra at $200 gives “10x Pro usage” every week and drops the five-hour limit for verified accounts. Max and Ultra also include Fable for up to half the weekly limit, where Pro pays for Fable with usage credits.
ChatGPT, per OpenAI’s Codex pricing docs and the Pro tiers article: Go is $8, Plus is $20, and Pro comes at $100, $200 or $500. Among the Pro plans, only Pro 500 includes Ultrafast mode. OpenAI also says Pro plans currently have no five-hour limit in Codex.
Weighing the next rung up on the Claude side? Claude Max vs Pro works through when the $100 plan pays off.
Usage limits: Claude vs ChatGPT
The two $20 plans count usage in different ways, and this is the row you feel every week.
Claude Pro has two limits. A session limit starts with your first message and resets in full five hours later. A weekly limit resets at a time tied to your account, shown in Settings, then Usage. Per Anthropic’s usage limits article, there is no fixed message count, and claude.ai, Claude Desktop and Claude Code all count toward the same limit. The pricing page puts Pro at “at least 5x more usage per 5-hour session than Free”. At a limit you can wait, upgrade, or turn on usage credits and keep working at standard API rates.
ChatGPT Plus splits chat from agent work. The pricing table lists everyday text chats as “Unlimited” on every plan, with a footnote that usage must be reasonable and follow OpenAI’s policies. OpenAI’s Plus article adds that Plus “may include usage limits such as message caps, especially during high demand.”
Agent work is metered on its own. Per OpenAI’s Codex help article, Codex, ChatGPT Work, ChatGPT for Excel and Workspace Agents share one allowance. On Plus, OpenAI estimates these local Codex messages per five hours:
- GPT-6.1 Sol: 15 to 160
- GPT-6 Luna: 350 to 3,000
Weekly limits may also apply, and cloud tasks use more of the allowance than local messages. Eligible Plus and Pro users can buy credits without changing plans.
So on Claude, one pool covers everything, and a long Claude Code session eats into the chat you were planning to have later. On ChatGPT, plain chat is listed as unlimited, and the coding agent and the Excel add-in share their own budget.
Every Claude plan’s caps and reset rules are in Claude usage limits, explained. The Codex side, plan by plan, is in Codex usage limits.
Coding: Claude Code on Pro vs Codex on Plus
Both $20 plans include a coding agent that edits your repo from the terminal, so a developer can pick either plan and still get one.
Claude Code on Pro. Claude’s pricing FAQ says Claude Code is included in all paid plans and shares the same usage limits as the rest of your plan. Per Claude Code’s model docs, the default model on Pro resolves to Opus 5.5, the same default as Max and Ultra. Fable is there too, paid with usage credits. Type /usage in Claude Code to see your plan’s usage bars. Our guide on how to check Claude Code usage covers the other ways to check.
Codex on Plus. OpenAI’s pricing docs describe the Plus allowance as enough to “power a few focused coding sessions each week”, with GPT-6.1 Sol and GPT-6 Luna. Plus also gets cloud integrations like automatic code review and Slack, and Codex Cloud for eligible accounts runs tasks on OpenAI’s computers while yours sleeps. Codex runs in the CLI, the IDE extension, on the web and in the ChatGPT desktop app. Type /status in a Codex session to see where you stand, or follow how to check Codex usage.
Which one codes better is a bigger question than the plan. Claude Code vs Codex compares them head to head, and Fable vs Opus covers when Claude’s top model is worth paying extra for.
One practical point for $20 buyers. On Pro, Anthropic publishes multiples, not counts, so you find your ceiling by using it. On Plus, OpenAI publishes a range, but a range of 15 to 160 is wide enough that you also find out by using it.
Office add-ins, files and connectors
For a business owner, this is about where the AI can reach your work.
Office add-ins. Claude Pro includes Claude for Microsoft 365, a set of add-ins for Excel, PowerPoint, Word and Outlook. Per Anthropic’s add-in docs, it reads and writes cells, formulas, pivot tables and charts in Excel, redlines Word documents with tracked changes, triages Outlook, and shares one conversation across the four apps. Free doesn’t get it.
ChatGPT Plus gets OpenAI’s extensions for Excel, Word, PowerPoint and Google Sheets, which the pricing table marks Limited on Free and Go. Per OpenAI’s spreadsheet add-in article, Plus use draws on the plan’s agentic usage limit, the same pool as Codex, and spreadsheet chats don’t have ChatGPT’s memory. The deeper guides are ChatGPT for Excel and Claude in Excel.
Making files. Claude can create .xlsx, .pptx, .docx and PDF files on every plan, Free included, per Anthropic’s file article. ChatGPT makes and edits Word, Excel and PowerPoint files, and Google Docs, Sheets and Slides when that app is connected, through ChatGPT Work. Plus gets Work on desktop, web and mobile, where Free and Go get a limited desktop version.
Connectors and MCP. MCP is the standard that lets an AI app use another tool, such as your CRM, your database or a board. Claude lets every plan add custom connectors to remote MCP servers, with Free limited to one, per Anthropic’s connector article. Those connectors can create, change and delete data, and they work on claude.ai, the desktop app and mobile.
ChatGPT Plus has plugins and developer mode (beta). But OpenAI’s custom MCP article says full MCP, including write actions, is for Business and Enterprise/Edu on ChatGPT web. Pro can connect custom MCP servers with read and fetch permissions only, and custom servers don’t work on mobile. Plus isn’t named in that article.
Projects. Both have them. Claude Pro has projects with no cap listed (Free gets up to 5). ChatGPT lists projects on every plan, and Plus bundles them with scheduled tasks and custom GPTs.
Images, voice, memory and context
Images. ChatGPT Plus has image generation, including image generation with Thinking. Claude’s pricing page lists none, and its file article describes data visualizations saved as PNG files. If you make social images, mockups or thumbnails, that one row can settle it.
Voice. Claude has voice mode on every plan. ChatGPT Plus has expanded voice and voice with video.
Memory. Claude remembers across conversations on every plan. ChatGPT Plus has expanded memory and memory with past chats.
Context window. This is how much the model can read at once. Claude’s table says up to 1M tokens, varying by model, on every plan. ChatGPT Plus lists 54K for GPT Instant and 256K for reasoning models, which OpenAI puts at about 320 pages of input. For a long contract, a big codebase or a year of notes, Claude reads more in one go.
Research and browsing. Claude Pro adds Research and Claude in Chrome. ChatGPT Plus has expanded deep research and a built-in browser.
ChatGPT Plus or Claude Pro: which to buy
Pick by the job you do most, then check the limit after a week.
- You make images, use voice a lot, or chat all day: ChatGPT Plus. Images and voice with video are on Plus, and plain text chat has no posted cap.
- You live in Word, Excel, PowerPoint or Outlook: Claude Pro. All four add-ins come with it, and Outlook isn’t on OpenAI’s add-in list.
- You want the AI to act on your own tools: Claude Pro. Its custom connectors can make changes from $20. On ChatGPT, full MCP starts at Business.
- You read long documents: Claude Pro, for the larger context window.
- You code most days: either. Claude Code on Opus 5.5 and Codex on GPT-6.1 Sol are both included. Codex adds cloud tasks. Claude Code shares its limit with your chat, so heavy coding leaves less for everything else.
- You want to save money over a year: Claude Pro at $200 a year works out to $17 a month. Plus has no yearly price.
- You want the cheapest way in: ChatGPT Go at $8, if its limited Codex and Work access is enough.
Which AI subscription is worth it? The one whose limit you hit less often for the work that earns you money. Measure it before you upgrade: /usage in Claude Code, Settings and Usage on claude.ai, and /status in Codex. If you hit the wall on both, the next rung is $100 on either side.
Paying for both: one workspace for Claude Code and Codex
Plenty of people run both. Claude for writing and Office work, ChatGPT for images and voice, and both coding agents on the same repo because two $20 plans go further than one. That is $40 a month, two five-hour windows and two weeks to track, and two agents whose work ends up in different places.
That’s the job Context Board does. Its AI usage widget shows what is left of each plan’s five-hour window and its week, Claude Code and Codex side by side, as rings that turn amber from 20% left and red under it, each with when it resets. Since Claude chat and Claude Code draw from one pool, the Claude Code ring is your Pro plan. The Codex ring is Codex’s allowance on your ChatGPT plan, not your ChatGPT chat.
- Claude Code moved 2 tasks to Done4m
- Codex added this week's signups1h
- Claude wrote the Friday summary2h
Claude is low this window. Batch jobs go to Codex until 3 PM.
A small runner on your computer reads what each app last heard from its own servers. You set it up once with one line from Connect, and only names, states and percentages leave your computer, never what was said in a session. You add AI usage yourself from the gallery, since it reads your own computer. The Agents template already has it.
Both agents work on the same boards. Claude Code and Codex add a board over MCP and can read it, send it data, add widgets, tasks and notes, and post updates. Claude on Pro can do the same through a custom connector. ChatGPT itself writes to a board only from Business and Enterprise/Edu workspaces. On Pro it can read your board, and the ChatGPT setup page has the details. For a board built around your agents, see the AI agent dashboard.
What Context Board doesn’t do: it doesn’t replace either subscription, add usage to either plan, or read ChatGPT’s chat limits. The board’s chat runs on your own Claude Code or Codex plan.
If you pay for both, start a Context Board and put both plans on one screen. Free for 3 days, then $12 a month.