fix(sync-model-stats): extend AA retry budget

[kilo-code-bot]

Summary

The sync-model-stats cron was reporting Artificial Analysis 500s to Sentry even when the AA API itself indicated transient retryable errors (e.g. Could not query the database for the schema cache. Retrying.). The previous retry budget of 10s only allowed a handful of attempts before fetchWithBackoff returned the failed response and the route surfaced it to Sentry.

This bumps the AA fetch budget to 2s base / 120s max, so we only get a Sentry report once retries are truly exhausted.

Verification

N/A — config-only change to retry timing.

Visual Changes

N/A

Reviewer Notes

The cron runs periodically; a 120s upper bound for the AA fetch is well within Vercel cron limits and only kicks in on sustained 5xx/429 responses.

[kilo-code-bot]

Code Review Summary

Status: No Issues Found | Recommendation: Merge

Files Reviewed (1 files)

• apps/web/src/lib/model-stats/sync-artificial-analysis.ts

---

_{Reviewed by gpt-5.5-2026-04-23 · 217,557 tokens}