Usage, not quality
Rankings count completed scoring jobs with a recorded model ID. They do not measure accuracy.
LIVE PRODUCTION USAGE
Job Playlist AI Model Usage Leaderboard
See which models scored jobs, their recorded volume, and provider reliability.
Recorded usage, not model quality
Completed scoring jobs—Saved successful queue completions
Named models on record—Across all recorded history
Provider attempt receipts—Recorded per-job attempts, not API calls
Provider split · Kilo / OpenRouter—Share of completed jobs
Model identity unresolved—Not assigned an invented model name
THIS WINDOW
Most-used scoring models
Ranked by recorded completed jobs.
Loading completed scores…
FROM PRODUCTION RECEIPTS
Scoring activity over time
Daily completed scoring jobs by model.
Last 30 UTC daysHover bars for daily details
Dates reflect queue-record updates, not necessarily provider completion times.
RECORDED MODEL ACTIVITY
Models used for scoring
Click a model for its evidence details or select two to compare.
Loading models…
Requested model ID, not verified upstream identity
| Rank | Model | Completed jobs | Share | Status | 7-day trend | Last used | Compare |
|---|---|---|---|---|---|---|---|
| Loading verified scoring data… | |||||||
Active: 24 hours. Recent: 7 days. Past: older. Trends compare consecutive 7-day periods.
SIDE BY SIDE
Compare two models
Compare recorded usage, not accuracy.
Choose two different models to compare their recorded production history.
Models scored different jobs, so this is not a quality comparison.
MODEL ROTATION
Rotation history
When each model was first and last used, and which models took over its work.
| Model | First used | Last used | Status | Replaced by |
|---|---|---|---|---|
| Loading… | ||||
OPERATIONAL RELIABILITY
Provider reliability
First-attempt success, retries and failures by class, per provider. Failure records identify the provider, not the model, so there are no per-model failure rates.
Loading provider receipts…
EVIDENCE
Sample sizes and history gaps
- Loading…
METHODOLOGY & LIMITS
What these numbers mean
Recorded model names
We show the selected model ID. Automatic routes without a confirmed model remain unresolved.
Provider reliability
Provider success uses recorded first attempts. Failures cannot reliably be assigned to individual models.
Not a quality score
Models scored different jobs. This page cannot determine which produces better matches.
Cost and speed
Reliable model-level cost and latency data is unavailable.
Historical activity
Counts include historical scoring activity, not just jobs currently open.