What causes "failed to get model statistics" in Wei-Shaw/sub2api?

Retry the request — transient DB failures are the most common cause

failed to get model statistics

Error message

failed to get model statistics

What it means

Returned while assembling the v2 dashboard snapshot when includeModels is true and h.getModelStatsCached(...) fails. The model statistics query uses usagestats.ModelSourceRequested plus the snapshot filters; the original error is masked by this generic message. Infrastructure-side failure.

Source

Thrown at backend/internal/handler/admin/dashboard_snapshot_v2_handler.go:214

	}

	if includeModels {
		models, _, err := h.getModelStatsCached(
			ctx,
			startTime,
			endTime,
			filters.UserID,
			filters.APIKeyID,
			filters.AccountID,
			filters.GroupID,
			usagestats.ModelSourceRequested,
			filters.RequestType,
			filters.Stream,
			filters.BillingType,
			filters.UpstreamModelMismatch,
		)
		if err != nil {
			return nil, errors.New("failed to get model statistics")
		}
		resp.Models = models
	}

	if includeGroups {
		groups, _, err := h.getGroupStatsCached(
			ctx,
			startTime,
			endTime,
			filters.UserID,
			filters.APIKeyID,
			filters.AccountID,
			filters.GroupID,
			filters.RequestType,
			filters.Stream,
			filters.BillingType,
			filters.UpstreamModelMismatch,
		)

View on GitHub (pinned to 073e92d171)

Solutions

Retry the request — transient DB failures are the most common cause
Shorten the date range or reduce filter cardinality (add model/user filters)
Inspect backend logs for the underlying model-stats error
Exclude the models section from the snapshot while investigating

Defensive patterns

Strategy: retry

Try / catch

// TS
try { models = (await getSnapshotV2({ includeModels: true, start, end })).models; }
catch (e) {
  if (e.message === 'failed to get model statistics') {
    await sleep(2000);
    models = (await getSnapshotV2({ includeModels: true, start: startPlus30d, end })).models; // narrower window
  } else throw e;
}

Prevention

Add filters (user, key, model) to shrink model-stats aggregations
Retry transient failures with backoff
Alert on repeated model-stats failures to catch missing indexes early

When it happens

Trigger: GET dashboard snapshot v2 with the models section included while the model-stats aggregation over the requested window fails (DB error/timeout).

Common situations: Deployments with many distinct model names making the model-stats GROUP BY expensive; database contention from concurrent snapshot builds; missing usage-stats indexes after migration.

Related errors

AI-assisted analysis of Wei-Shaw/sub2api@073e92d171 (2026-08-15). Data as JSON: /api/errors/06bf65afc154b76f. Report an issue: GitHub.