Skip to main content

New Features

Model API Cost Observability Center

We are introducing a Cost Observability Center for Novita Model API, transforming billing from a “black-box invoice” into an attributable and controllable cost visibility system, directly addressing one of the most critical enterprise pain points — not knowing where the money is spent. Key highlights:
  • Three-level Cost Drill-down
    We now support hierarchical cost exploration from organization → team → member → API key. Strict permission isolation ensures secure and non-overlapping data access across roles.
  • Budget & Quota Management
    We introduce a Budget Quota card across User and API Key levels, showing real-time usage progress. This enables precise cost control and predictable budget management across teams and projects.
  • Cost Trend Visualization
    A new cost trend dashboard provides structured visibility into spending patterns over time, with support for daily, monthly, and custom time ranges. It offers clear insights into usage distribution across different models.
  • Cost Ranking & Hotspot Detection
    Two new ranking tables — Top Models by Cost and Top API Keys by Cost — help instantly identify cost drivers and usage hotspots.

Model API Request Logging

We now support detailed request-level logging for Model API calls, significantly improving debugging efficiency and operational transparency for production AI systems. Key highlights:
  • Full request and response traceability
  • Searchable logs by request ID, model, or timestamp
  • Error diagnostics and debugging support
  • Latency tracking per request
  • Secure and structured log storage for enterprise use

Improvements

Enhanced Model Marketplace Experience

We’ve significantly improved the Model Marketplace to make model discovery and selection easier and more intuitive. These enhancements help developers quickly find the right model for their use case and reduce evaluation time. Key improvements:
  • Improved filtering and categorization
  • Enhanced model cards with richer specifications and benchmarks
  • Faster browsing performance and smoother navigation
  • Better model discoverability for new and trending options
Last modified on June 25, 2026