Tests run on your machines, not on this server. The console authorises runs,
meters your plan and keeps your history. Prompts, responses and API keys never
leave your infrastructure — and your measurements never include our network's
noise.
Get started: build your first test in the console, then run it on your own
machine — no test has synced to this account yet.
You requested — .
You're on for now while that gets set up.
This month
–runs started this month
Plan limits
Running now
Live tests executing on your machines, reported by their agents.
Usage history
Runs per month; green shows how many met their SLAs.
Billing
Update your payment method, change seats, or cancel your subscription.
Danger zone
Deletes your account. You'll be signed out immediately and won't be
able to sign back in — contact support if you need it restored.
Run history
Runs synced from your agents with --sync.
Run detail
Build a test here, then run it on the machine you want to measure. The console
checks it against your plan and hands you a ready-to-run config.
Test
Your key stays on your machine — the config only names
the environment variable.
Use a prompt that reflects your real workload — token
counts and latency depend heavily on it, so a generic prompt gives you
generic numbers. Multi-line is fine.
Leave blank for no limit. For gpt-5/gpt-6/o1/o3/o4
models, reasoning tokens count toward this limit too.
SLA & cost
Grades response content, not just speed — but only once the
downloaded config has quality.checks configured, which this form
doesn't build. Setting a value here adds a starter quality: block
with instructions to the generated config.
Run it on your machine
0 — sign this machine in once (skip if you already have)
1 — download the config, or copy it
and save it as infergauge.yaml
2 — run it; results appear here automatically
Connect a machine
Each machine that runs tests signs in once.
Agent tokens
Each token identifies one machine or pipeline. Revoke to cut it off.