πΈ Version 3.12.0
β Back to the Main Changelog
February 11, 2026
Summaryβ
Major improvements to metrics collection, cost tracking, usage logging, and dashboard charts. Fixed Chinese Traditional translation corruption, bot mirroring issues, and emoji handling in the free engine. Added personal usage stats for users, bot translation admin controls, and background metrics collection. Fixed dispute evidence lookups and improved the admin dashboard's queue throughput graphs.
New Featuresβ
Personal Usage Statsβ
- Your Translation Stats - Every user can now see their personal translation recap on the User Settings page, including total translations, characters translated, tokens used, estimated cost covered, a 30-day summary, and a breakdown by feature (auto-translate, conversation bridge, right-click translate, etc.)
Bot Translation Admin Controlsβ
- Bot & Webhook Toggles - Admins can now enable or disable bot message translation and webhook message translation directly from the admin settings panel, giving full control over which automated messages get translated in mirrors
Background Metrics Collectionβ
- Always-On Tracking - Queue and system metrics are now collected automatically in the background, even when no one is viewing the admin dashboard. Historical data is always available when you open the dashboard.
Improvementsβ
Realistic Cost Estimationβ
- Infrastructure Cost Tracking - Free-tier translations now include estimated infrastructure costs to reflect actual server expenses
- Premium AI Cost Tracking - Improved cost accuracy for premium translation engines
- Adaptive Cost Display - The dashboard now shows costs with enough precision to be meaningful, even for very small amounts
Dispute Evidence Lookupβ
- Complete Activity Tracking - The lookup now captures all user activity including free-tier usage (conversation bridge, auto-translate, mirrors) that was previously invisible
- All Action Types Visible - All translation events now appear in the evidence log, not just premium and login events
Bot Queue Throttlingβ
- Separate Bot Queue Speeds - Bot and webhook translations now process at a slower rate than user translations, preventing automated messages from hogging the translation queue and keeping user translations fast
Dashboard Chartsβ
- Fixed Time Range Presets - All time range options (1h, 6h, 24h, 7d, 30d, MTD, QTD, YTD) now work correctly. Previously, longer ranges silently defaulted to 24 hours.
- Smart Time Bucketing - Data is now aggregated into appropriate time buckets based on the selected range (1-min for 1h, 15-min for 24h, 1-hour for 7d, etc.) to keep charts readable with ~60-120 data points
- Improved X-Axis Labels - Chart labels now use context-appropriate formatting and are angled for readability
- Default to 1 Hour - Charts now default to the 1-hour view for live monitoring instead of 24 hours
Bug Fixesβ
- Fixed Chinese Traditional translations producing corrupted/repeated characters (e.g. "ιιιι") when using the free engine. Traditional Chinese now translates correctly for both
zh-Hantand legacyztlanguage codes - Fixed bot and webhook messages (e.g. Carl-bot posts) not being translated in mirror channels when bot translation was enabled
- Fixed emoji handling in the free translation engine -- messages containing Unicode emojis (e.g. "εθδΊΊζΈδΈε€ π") no longer produce garbled translations. Emojis are now preserved through translation untouched
- Fixed auto-translate and conversation bridge events not being logged, making it impossible to track those translation types in usage history
- Fixed mirror translation activity being attributed to "system" instead of the actual user who sent the message, so usage logs now correctly show who triggered each translation
- Fixed dashboard charts ignoring time range selection, causing all ranges beyond 6 hours to show identical data
- Fixed dispute evidence only showing paid-tier events, missing all free-tier translation activity
- Fixed cost estimation showing $0.0000 for all free-tier translations