The two steps

Shrink The Prompt. Then Publish.

Tap a point to read about it. The bars show the shape of the idea, not measured numbers.

One. The reasoning model

It Thinks Well. It Writes Plainly.

I use reasoning models, at very low cost,
to distill my prompts into smaller prompts.

In my experience they think well but write plainly.
They play chess. They can even build a chess game.

So I let them do the thinking.
I do not ask them for the polish.

Two. The publishing model

It Has The Polish

Publishing models, like Claude Fable,
render the finished output.

They carry a lot of content,
and a lot of refinement from users.

That polish is why they cost more.

Three. Why it saves

The Expensive Model Reads Less

The polished model reads the small prompt,
not the big one. Fewer tokens at the high price.

We count tokens for every query.
So the difference shows up in the numbers.

This is a big distinction.
We choose the model for each job.

Open Earthgrid.ai The private island Stop chatting. Start building.

Bring it to your stage

Try It With Your Own Prompts

I show it live, with a prompt from the room.
We shrink it, then publish it.

Invite me to your stage Hear the talk All talks