Every LLM call your agent makes — what it cost, whether it cached, and why. Per call, per session, in real time.
Join the Beta — FreeOpenClaw is powerful, but it doesn't tell you where your money goes. Most users are overpaying without knowing it.
You told your agent to check something daily three weeks ago. It's still running. Every heartbeat, every cron job, every background task costs money — and you forgot it exists.
You see a monthly bill, not what each call cost. Was that cron job $0.02 or $2.00? Did caching actually work? No way to know.
Your agent uses Sonnet to respond "OK" and Opus to run a sleep command. You're paying premium prices for tasks any model could handle.
No SDK, no code changes. Run a single npx command and you're live.
npx stairwell init
Configures OpenClaw automatically in seconds. Your costs appear on the dashboard in real time.
Free for beta users. Full cost visibility from day one.
Every LLM call appears on your dashboard the moment it happens. See the user message, model response, tokens, cache behavior, and cost — all in one live feed.
Not just a monthly total — see exactly what each API call costs in cents. Know immediately when something is expensive and why.
See cache read/write tokens on every call. Know if Anthropic's prompt caching is actually saving you money or if your cache is constantly invalidating.
Works with whatever provider you use. One dashboard for all your models — Claude, GPT, Gemini, MiniMax, DeepSeek, and everything on OpenRouter.
Stairwell classifies every call and routes simple ones to cheaper models. Complex calls stay on your primary model, untouched. Beta users get smart routing first.
Your agent makes an LLM call through Stairwell like normal.
"Thanks, got it" goes to Haiku.
"Why is my wife mad at me?" stays on Opus.
Simple calls cost 90% less. You never notice the difference.
Join the beta and get full cost visibility on your OpenClaw agent — free.
Join the Beta — Free