System status
A site that hides its errors cannot ask you to trust it. This page shows which collector jobs ran, which failed, and how much time JEV calls take.
System healthStale 16d
Model: preliminary, trained 396 hours ago, retrained daily: no (last retrain Oct 02, 23:02 UTC). Model card
Median JEV time over the last hour
-
all JEV calls together, not one headline
Slow JEV calls, last hour (p95)
-
19 of 20 calls are faster
Collector jobs running
8 of 10
jobs of the collector, not data sources
JEV calls, last hour
1,346
1,346 errors (100.0%)
JEV calls, 24 h
32,205
32,205 errors, the last 24 h
Provider
official
the primary JEV endpoint
Daily cap used
64%
of 50,000 calls
Data freshness
| Section | State | Age | Limit |
|---|---|---|---|
| Coin prices | Stale 16d | 16d 1h | 5m |
| Market overview | Stale 16d | 16d 1h | 5m |
| News | Stale 16d | 16d 1h | 15m |
| Stablecoins | Stale 16d | 16d 1h | 30m |
| Memecoin radar | Stale 16d | 16d 1h | 15m |
| NFTs | 14m ago | 14m | 2h |
| Signals | Stale 16d | 16d 1h | 15m |
| Scoreboard | Stale 16d | 16d 1h | 15m |
| System health | Stale 16d | 16d 1h | 5m |
| Token list | Stale 16d | 16d 1h | 15m |
| Price history | 5h ago | 5h 47m | 8h |
Collector jobs
Stale 16d| Job | State | Last run | Fails, 24 h | ||
|---|---|---|---|---|---|
| dex | Failing | 719 | |||
| dex: DexRunError: JEV failed: Every JEV call failed for dex. First error: official JEV call failed: https://api.typesafe.ai/v1/systemone returned HTTP 402: {"detail":{"error_type":"billing_error","message":"Your organization has no available TypeSafe API credits. Please add more credits and/or set up auto-reload at https://console.typesafe.ai/settings/billing"}} | |||||
| growth | OK | 0 | |||
| health | OK | 0 | |||
| history | OK | 0 | |||
| market | OK | 2 | |||
| news | Failing | 288 | |||
| news: JevError: Every JEV call failed for news. First error: official JEV call failed: https://api.typesafe.ai/v1/systemone returned HTTP 402: {"detail":{"error_type":"billing_error","message":"Your organization has no available TypeSafe API credits. Please add more credits and/or set up auto-reload at https://console.typesafe.ai/settings/billing"}} | |||||
| nfts | OK | 0 | |||
| retrain | OK | 0 | |||
| stablecoins | OK | 0 | |||
| tracker | OK | 0 | |||
How good are JEV's ratings here?
Full comparisonImpact: agreement with a reference model
70.0% (64.2% to 75.8%)
constant guess 57.9%, keyword rules 40.0%
Sentiment against keyword rules
+3.3 points (-5.0 to +11.7)
Not clearly better than the keyword rules: the interval includes 0.
When JEV says routine, the reference agrees
52.5%
precision of the routine class
Agreement with another model, not people. 240 items, 7 days, English headlines, measured Sep 29, 10:31 UTC. Read the intervals, the baselines, and the limits.
Local models against the official one
JEV runs through the official TypeSafe API. We also operate two JEV models on our own GPUs to compare against it. That comparison is offline right now. This page shows agreement only when comparison samples exist.
No comparison samples: the local models are offline.