A free bench of tools for getting more out of AI models — count tokens, plan caching, route models, cut waste, wherever you build.
Everything inside the bench, and every thread tying it to the rest of the platform. Click any piece to read what it does.
Workbench is TurnTwo's free toolkit for getting more out of AI models. It answers the questions you can't eyeball — how many tokens is this, what will it cost, which model fits, where is a prompt wasting money — with tools that run right in the browser, no signup for the free ones.
Most tools are pure logic and free. A few add an optional step that calls a model, and those run on your own key — never the owner's. It's a way to meet the practice, not a paywalled product.
Around twenty tools, grouped by the job they do. The diagram above walks the whole bench — this is the plain-language version.
This is the layer a normal encyclopedia never gives you: not just what Workbench is, but the decisions that shaped it and the reasons behind each one.
The mechanism, told the way a public exhibit is allowed to tell it. Where a detail crosses into spend or provider specifics, the trail stops at a seam — that material lives only in the private brain.
The candid part. A wiki earns trust by being clear about what isn't here yet — this is the layer that makes it a true reference rather than a brochure.