Media Usage & Cost Visibility in the Dashboard
The dashboard now goes beyond chat tokens and surfaces what you spend on speech, OCR, and image generation — plus filtering and stability improvements.
✨ New Features
- Media usage cards: The overview now includes dedicated cards for speech-to-text (Whisper), text-to-speech, OCR, and image generation. Each shows the usage volume and its cost in EUR, so non-chat spend is visible at a glance.
- Media models in the provider breakdown: The usage-by-provider donut now includes media models — Whisper, TTS, Mistral OCR, DALL·E, Flux, and gpt-image-1 — alongside your chat models, for a complete picture per provider.
- Token-type filters: Statistics can now be filtered by token type, making it easier to drill into exactly the usage you care about.
🔧 Improvements
- Broader usage coverage in tables: Usage tables now account for speech, OCR, and image-generation activity, not just chat.
- More reliable loading: The dashboard handles large date ranges far better and no longer crashes on heavy queries.
🐛 Fixes
- Corrected an end-of-day boundary so daily totals include the full day’s activity.