Playground
Run a simulated agent task. No signup.
Four pre-scripted scenarios. Click run. The timeline, cost, and cache-hit rate behave the way the real system does. Numbers come from real evals, timing is deterministic.
Warm run
Storylines are deterministic. Editing the prompt doesn't change behavior in simulated mode.
Run timeline
- 1model_call · qwen3-4b · 58ms · $0.0150missworker-1
Cache hit rateidle
0.00
Tokens
312
Cost
$0.0150
Budget$0.01 / $2.00
How this works
Simulated mode runs entirely in your browser with no backend calls. Timing, cache hits, and costs come from fixed scripts so every visitor sees the same story. Switch to Live mode to run real tasks against your own KovaServe cluster.