Observed inference events
0
Across stored probe history.
Tokenmill / Live ledger
LLM inference endpoints, measured as a living surface rather than a static list.
Current probe results, account gates, free-tier signals, reset horizons, and immediate next moves. The goal is not catalog sprawl. The goal is operational clarity.
Providers tracked
0
Live now
0
Credential-gated
0
Worth getting
0
Coverage
Probe proportions
Waiting for probe data.
Usage widget
Token, quota, and latency history
Observed total tokens
0
Only from providers that returned usage metadata.
Mean latency
0 ms
Across successful live probes with latency data.
Daily token burn
Visible pressure against quotas over time
No pressure seen yet.
Quota posture
Unknown
Waiting for enough history to classify pressure.
Inference trigger rate
0/day
Based on successful live probes in stored history.
Fastest live lane
None
Need more live providers before real routing decisions stabilize.
Reset horizons
Quota and cadence framing
Credential Ramp
What to get next, and how hard it is
Provider ledger
One by one, with blockers spelled out
Provider History
Who is actually burning tokens, and how hard?
Smooth operation
Fallback stack and sufficiency guidance
Guarantee Plan
What would actually be enough
Raw detail
Reliable table
| Provider | Status | Access | Horizon | Worth | Latency | Blocker / note |
|---|
Provenance