Playground
Every model. Everything it can do. One account.
Models from several providers, in one list, each with what the provider built in: long context, reading images, and in some cases writing files for you. One login, from wherever you sit, and one wallet that pays for all of it.
Illustrative session. Answers and prices are samples.
Many providers. Many logins. This room has one.
Most teams that use these models for real work end up with a drawer of logins. One provider for writing, another for code, a third that is better at reading a long PDF. Each wants its own signup, its own card, and its own rules about which countries it serves. Half the team has one of them. Nobody has all three.
The Playground is one login for all of them. Sign in, pick the model from thecatalog your administrator switched on, and use it the way its provider built it.
For an administrator it is also one roster, one allowance per member, and one ledger. Not a set of invoices from companies you never chose.
One account
Every model your admin allowsone list
Each at full capabilityas shipped
Wherever you sign insame list
One walletevery model
Scroll, and the drawer closes.
Each one, the way its provider built it.
Hover a provider. Nothing on this grid is trimmed to make the list look tidy; a lit lamp means the capability works as the provider ships it.
Pick a model
The list holds every model your administrator allowed, across providers. Switch mid-thread and the conversation is replayed to the new one, so you can see how a second model handles the same brief.
Compare three
Send one message to up to three models side by side. Each column carries its own answer, which usually ends the argument about which one to use.
Set the brief
A system prompt and a temperature per conversation. Tell it to answer as a terse financial analyst, in French, in tables. If your administrator locked the sampling settings, the page says so and uses theirs.
Keep the thread
Conversations are saved and searchable, and you can rename them. Export the whole thread when it becomes a document worth keeping, or delete it when it does not.
What goes in, and what comes out.
This is where the full-capability point stops being abstract. Documents are read as text and counted before you send. Long documents are fine; the counter tells you how much of the context they take. Images go to the models that can see them.
Models marked as able to create files hand back real documents. Any answer can also be exported in the format the next person needs, and any table in it as a sheet.
reads it
Word
A report goes in as text. A rebuilt spreadsheet comes out as a file.
Every answer says what it cost.
Cost isn’t the point of this page, but you’ll want to know. Under each reply: the tokens in, the tokens out, and the price in dollars. In a comparison, each column carries its own. The sidebar shows what you have spent this month against the allowance your administrator set, if they set one.
There is no seat fee for the Playground. Usage is metered in tokens at the model’s own rate, and each run also lands in the organization’s ledger with your name, the model, and the exact cost.
Length of the answer
0tokens out
$0.0000
Sample rate. The real one is under the answer.
Who decides what
Same list, same wallet, from wherever you sign in.
A hotel room in a city where one of the providers does not sell accounts. The list is the same.
One login
No provider accounts to open, no personal cards, no personal email on a company tool.
Same roster
The models your administrator switched on travel with you. So does the allowance.
Same ledger
A run from a hotel room lands in the same ledger as a run from your desk.
Six things you can check.
Checked against the product on August 23, 2026. If one of these stops being true, the page changes.
The Playground comes with every account.
Open an account, add money to the wallet, and every model you switch on is there, from wherever you’re working.
No subscription. No seat fee. Money in a wallet, spent on tokens, storage and data, and your balance never expires.