Comparison

Busabase vs ChatGPT Space

ChatGPT Space is OpenAI's shared home for pages, slides, files, and Sites, where your team writes together with ChatGPT, Codex, and dots. Busabase is the open-source workspace where any agent — Claude Code, Codex, Cursor, ChatGPT, or any MCP client — writes your team's structured data, docs, and skills, and every change arrives as a Change Request you can read, review, and undo. Space is built around documents people write with OpenAI's AI. Busabase is built around the data every agent you use reads and writes.

OpenAI added Space to ChatGPT on 29 September 2026. This page compares the two by the jobs teams bring to them, and says plainly where Space is the better choice.

Last verified 13 November 2026 against OpenAI's ChatGPT Space documentation, help center, release notes, and pricing page, and Busabase's current product. Where OpenAI's pages do not describe something, we say “not described”, which is not the same as “cannot”.

Connect your agent All comparisons

The short verdict

Choose Busabase when

agents from more than one vendor write data your team and other agents rely on — customer records, research, content, operating docs, skills — and you want every change recorded with a diff, its author, and undo, on infrastructure you can choose, at a flat price per seat.

Choose ChatGPT Space when

your team already works in ChatGPT and the deliverable is a document, a deck, or a small Site written together with ChatGPT, Codex, and dots, with nothing to host.

Use both when

Space holds the team's drafts and discussions and Busabase holds the records agents produce. ChatGPT connects to Busabase over MCP, so a conclusion reached in Space can land in Busabase as a Change Request.

The core difference: every change is a Change Request

Think of it as a pull request for your data. In Busabase, every edit — a record, a Doc, a Skill file, a schema change — is a Change Request: a branch with a chain of immutable commits. Merging moves the record forward; nothing is overwritten.

ChatGPT Space reviews AI text where it is written. OpenAI's Pages guide says you can review text ChatGPT generates “in place. You can accept it, ask for a change, or undo it,” and “Show attribution” shows who contributed, when available. That is a good answer for a paragraph the reader is looking at. OpenAI's Space documentation does not describe field-level diffs, proposals that wait before they apply, or version history for what agents write — and the next reader of a record is usually another agent, not a person.

A message and a diff on every write. Each change says what it does and shows the field-level before and after.
Know which agent did it. The author, and the key or tool the change came through — Claude Code, Codex, Cursor, ChatGPT — stay on the commit.
Undo is built in. Every record, Doc, and Skill keeps its full history, and any change can be rolled back.
Review when you choose. Writes with write access land immediately. Give an agent a proposal-only key and its changes wait in the Inbox for a person to approve or send back.

A Busabase Change Request under review: its message, a before-and-after diff of a record's body, and Approve or Request changes controls — A Change Request in the Inbox: what changed, who proposed it and through which tool, and the before and after of the field, with Approve or Request changes.

The review lives on the data, not inside one vendor's editor — so it still works when three different agents write to the same Base.

What else Busabase gives your agents and your team

Space is organised around pages and files. Busabase adds the parts teams need once agents produce data, not just documents.

Typed Bases with views

Records with typed fields that agents query and update. Every Base opens as a table, gallery, kanban board, calendar, or Gantt chart, with saved views and AND/OR filters.

A team knowledge base

Docs, files, and Bases sit in one tree, shared by people and agents, with roles for members and keys for agents.

Grep across everything

One regex search covers files, Doc bodies, Base records, and agent prompts, and returns the line and its context — so an agent finds the exact evidence before it reads further.

Skills every agent shares

Upload your team's Skills once. Claude Code, Codex, Cursor, and every other connected agent read the same version, and editing a Skill is a Change Request like any other.

Apps with four engines

Ask your agent to build an AirApp on your data. Run it in the browser, on your own machine, in a cloud sandbox, or as a website that stays online.

Websites on your domain

The Website engine runs on Cloudflare Workers, can have its own backend endpoints, and takes your own domain. Your agent can preview and publish it through the CLI, API, or MCP.

Any agent, not one vendor's

MCP, a CLI, an OpenAPI REST API, webhooks, a Python SDK, and plugins for Claude Code and Codex. ChatGPT connects too, over MCP.

Open source, runs where your data must

MIT-licensed. Run it on your laptop, on your own servers, fully offline, or on Busabase Cloud. A space exports to plain files: Markdown, Base definitions, and JSON records.

Flat per-seat pricing

Free for 3 seats with every node type, then $10 or $20 per seat per month billed yearly. No metered overage, and self-hosting is free with no seat limit.

A Busabase Base showing agent-written records as a table with typed columns and saved views — A Base in Busabase: records an agent wrote, with typed fields, readable by anyone on the team. Saved views filter the same records.

Comparison by the job you are trying to do

Rows are the jobs teams bring to these products, not feature checkboxes. Where ChatGPT Space is the right tool, the table says so.

What you are trying to do	Busabase	ChatGPT Space
See every change before and after, and who made it	Built in — a Change Request with a field-level diff, the author and the key it came through, full history, and undo	AI text on a Page can be accepted, changed, or undone in place; attribution when available. Field-level diffs and version history are not described
Let one agent propose changes without applying them	Yes — a proposal-only key; its changes wait in the Inbox for a person	Not described for Space content. A dot can bring results back for your review
Let agents from any vendor read and write the same data	Yes — Claude Code, Codex, Cursor, ChatGPT, or any MCP client	ChatGPT, Codex, and dots work in Space; access for agents outside OpenAI is not described
Keep structured records with typed fields and views	Yes — Bases with table, gallery, kanban, calendar, and Gantt views	Spreadsheets, and tables inside Pages. A shared typed record store with views is not described
Run a team knowledge base	Yes — Docs, files, and Bases in one tree, with member roles and agent keys	Yes — pages, slides, files, and Sites grouped into spaces and shared with your team
Find exact text across everything	Yes — one regex grep across files, Doc bodies, records, and agent prompts	Ask ChatGPT about your files; an exact-match search across Space is not described
Give every agent the same team Skills	Yes — upload once; every connected agent reads the same version, and edits are Change Requests	ChatGPT has Skills; sharing them with agents outside ChatGPT is not described
Build apps and websites on your data	Yes — AirApps in the browser, on your machine, in a cloud sandbox, or as a website on your own domain	Sites builds and hosts websites, with database storage and custom domains, hosted by OpenAI
Connect over MCP, CLI, or a REST API	Yes — MCP, CLI, OpenAPI, webhooks, Python SDK, and agent plugins	Space connects outward to tools such as Google Drive and Slack; an API or MCP route into Space is not described
Keep the data out of a single vendor	Yes — MIT open source, self-hosted or fully offline, exports to plain files	Hosted by OpenAI; no self-hosted option is described
Predictable pricing as the team grows	Flat per seat — free for 3 seats, then $10 or $20 per seat, no metered overage	Space is on Pro (from $100 a month), Business ($20 to $100 per seat), and Enterprise; not on Free, Go, or Plus
Write documents and slides with AI, co-edited live in ChatGPT	Not its job — Busabase has Docs, but not Space's live co-writing with AI	Yes — this is Space's home ground

Coming next

Three features are being built now. They are not available yet; we list them so you can plan, and this page will be updated when they ship.

App users In development

People who use the apps you publish — a shop's customers, say — sign in with an email code and see only their own records. The platform enforces it, and they never become workspace members or take a seat.

Ontology In development

Define the business objects your team talks about — Company, Contact, Deal — and the links between them on top of your existing Bases, so people and agents share one vocabulary. Changes to it go through Change Requests too.

Fine-grained permissions In development

Row- and field-level access inside a Base, scoped agent grants with an expiry, and access requests a person approves — so each agent gets exactly the access its job needs.

Reviewing a paragraph is not the same as reviewing a record

Space answers a real question well: how does a team write the same document with AI without stepping on each other? OpenAI's Pages guide says you can share a page “with edit access to write together in real time,” review generated text in place, and mention ChatGPT in a comment. Its Space page adds that you can “tag ChatGPT, Codex, or your dot to make changes or take on next steps.”

Accept-or-undo works because the person reading the paragraph is the person reviewing it, at the moment it is written. A field in a record is different. The next reader is usually an agent, a report, or an app, and the review has to survive as part of the record: what the value was, what it became, who proposed it, and who accepted it.

There is a second difference. Space's agents are OpenAI's — ChatGPT, Codex, and dots — working inside OpenAI's product, which is what makes its editing experience coherent. Most teams' agents are not all from one vendor: Claude Code in one repository, Cursor in another, Codex on a schedule, ChatGPT for research. When they all write to the same data, the review cannot live in any one of their editors. In Busabase it sits on the write path of the data and stays attached to the record afterwards.

Using them together

For many teams the useful arrangement is both:

People and dots draft, discuss, and present in Space: briefs, meeting notes, decks.
When a conclusion has to become a record — a supplier decision, an account update, a price — ChatGPT writes it into Busabase over MCP.
Give ChatGPT a proposal-only key and the change waits in the Inbox, with its field-level diff, until the owner approves it.
Claude Code, Cursor, Codex, and your AirApps read the accepted value from the same Base.

Space keeps the team's writing fast; Busabase keeps the data every agent relies on recorded and reversible. See how to connect ChatGPT to Busabase.

Pricing at a glance

Busabase

Free: 1 space, 3 seats, 5,000 records, 1 GB of files, every node type including Skills, workflows, and AirApps.
Plus: $10 per seat per month (billed yearly). Pro: $20. Enterprise: custom.
Flat per seat, no metered overage. Self-hosting is free under MIT with no seat limit.

ChatGPT Space

Included with Pro (from $100 a month, for individuals), Business ($20 per Standard seat or $100 per Premium seat per month billed annually; $25 or $125 monthly), and Enterprise (contact sales). Not on Free, Go, or Plus.
Dots come with Pro and Business Premium seats. Sites hosting above the monthly allowance is billed from 2027.

Prices as published on 13 November 2026. See Busabase pricing and ChatGPT pricing for current details.

Where ChatGPT Space is the better choice

Your team already works in ChatGPT. If people open ChatGPT every morning, a shared space inside it is easier to adopt than any tool they have to remember to visit.
The deliverable is a document, a deck, or a small Site. Live co-editing with ChatGPT, accept-or-undo on AI text, comments that mention ChatGPT, and slides you can present or export — Busabase's Docs are not a peer for this.
You want an agent that goes and does the next task. A dot has its own cloud computer and keeps working between conversations. Busabase has no built-in agent; you bring yours.

Busabase's limits, plainly: no live co-editing suite, no presentations, and no built-in agent. It has no realtime subscriptions and no vector index. Sign-in for the end users of the apps you publish is in development (App users) but not available today.

Frequently asked questions

Does ChatGPT Space have a review step for AI changes?

Yes, for text: AI-generated text on a Page can be accepted, changed, or undone in place, and attribution shows who contributed when available. OpenAI's Space documentation does not describe field-level diffs, proposals that wait before applying, or version history for structured data.

Can Claude Code or Cursor write into ChatGPT Space?

OpenAI has not described a public API or MCP route into Space for agents outside its own. Space is built around ChatGPT, Codex, and dots. In Busabase, every MCP client, the CLI, and the REST API reach the same data.

How is a Change Request different from accepting AI text?

Accepting text is a decision a reader makes about a paragraph in front of them. A Change Request is the change itself, kept on the record: the message, the field-level diff, the author and the tool it came through, and the history — so it can wait for review and be rolled back later.

Can ChatGPT write into Busabase?

Yes, over MCP, within the permissions of the key you give it. With a proposal-only key its writes wait as Change Requests for a person to review. See the ChatGPT setup guide.

Does every agent write in Busabase need approval?

No. Writes from anyone with write access land immediately and keep their history. Review is opt-in: give an agent a proposal-only key, or ask it to submit a change for review.

Can I self-host ChatGPT Space or Busabase?

OpenAI describes no self-hosted option for Space. Busabase is MIT-licensed and runs on a laptop, your own servers, or a fully offline network. See the self-hosting guide.

Next step

Connect an agent you already use — ChatGPT, Claude Code, Codex, or Cursor — and ask it to put one real piece of work into a Base. Then open the Change Request it created and see exactly what changed.

Connect your agent Browse templates

Sources: ChatGPT Space, Space (ChatGPT Learn), Pages in Space, ChatGPT Space: sharing, data, and controls, ChatGPT Space for Enterprise, ChatGPT release notes (29 September 2026), ChatGPT Sites, ChatGPT pricing, Busabase pricing