ccflare -> better-ccflare

Automatically distributes requests across multiple Claude accounts with session-based tracking and auto-fallback to prevent rate limit errors, as detailed in the load balancing features.

Supports Claude OAuth, AWS Bedrock, Vertex AI, OpenRouter, and other providers with universal API format conversion, enabling seamless integration across AI services.

Provides a web dashboard with token usage, latency, and cost tracking per request, including batch processing and advanced filtering for operational visibility.

Includes automatic failover, configurable retry logic, SQLite/PostgreSQL persistence, and security hardening for reliable deployments in high-throughput environments.

Claims <10ms overhead with lazy loading and request deduplication, ensuring low latency for API proxying without significant impact.