Most AI tools give you a response and you take it or leave it. You have no idea how it got there. The Playground changes that. Now you can watch the model reason through your prompt step by step, see exactly how fast it's generating, and iterate with keyboard shortcuts that make the loop feel instant.

What You See Now

When you send a prompt in the Playground, the response area splits into two phases. First, the Thinking phase - the model's chain-of-thought reasoning appears in real time. You can see it consider different approaches, weigh trade-offs, and build its answer piece by piece.

Then the Generating phase begins and your final response streams in. Two sections. One conversation. The toggle between them is always visible so you can check the reasoning behind any response.

This isn't a summary of what happened. It's the actual internal reasoning, streaming as the model produces it. You see every consideration, every discarded approach, every decision point.

Live Stats

The Playground tracks three things while the model works.

Characters per second

A real-time throughput counter showing how fast the model is generating. Useful for comparing models, spotting bottlenecks, and understanding performance.

Phase indicator

Shows whether the model is in Thinking or Generating mode. You always know which part of the process you're watching.

Syntax highlighting

Code blocks in the response are highlighted by language automatically. Python, TypeScript, JSON, SQL - whatever the model writes, it's readable.

Keyboard Shortcuts

Two shortcuts. That's all you need.

Cmd+Enter (or Ctrl+Enter) - Submit

Send your prompt without reaching for the button. Write your prompt, hit the shortcut, and the model starts working. No mouse needed.

Escape - Stop generation

Kill the current response if the model goes off-track. You don't have to wait for it to finish. Stop it, refine your prompt, submit again.

The loop should feel like a conversation, not a form submission. Submit, read, stop if it's wrong, refine, submit again. These shortcuts make that loop fast enough that you stay in flow instead of waiting.

Why This Matters

Transparency changes how you prompt

Chain-of-thought transparency is becoming the standard for anyone building with AI. When you can see how the model reasons, you write better prompts. You catch when it goes off-track early instead of after it finishes. You learn what the model handles well and where it struggles. The Playground is the starting point for ASI:One's developer experience - from experimenting here to building your own agents with the uAgents framework.

Better prompts Seeing the reasoning teaches you what instructions the model responds to and what it ignores.
Faster debugging When the output is wrong, the thinking section shows you where the model's logic broke down.
Model comparison Characters per second and reasoning quality give you concrete data to compare models, not just vibes.

Try It

Three steps. Under a minute.

Go to the Playground

Open the Playground from the sidebar or navigate directly.

Enter a complex prompt

Try something that requires reasoning - "Plan a weekend trip to Austin for two people who love live music and tacos" or "Write a Python function that merges two sorted lists without using built-in sort."

Watch the Thinking section

See how the model breaks down the problem before generating the answer. Toggle between Thinking and Generating to compare the reasoning with the final output.

Keep Reading

Product Update

See how your AI reasons

Open the Playground and watch the model think through your next question. Real-time reasoning. Live stats. Keyboard shortcuts.

Open Playground

Watch Your AI Think