January 2026: Security, Billing, CLI Webhooks, and Observability

January was a broad reliability and product maturity month. We strengthened account security, improved billing visibility, added local webhook development support, and made request logs more useful for real debugging.

Two-factor authentication

We added two-factor authentication flows for sign-in and organization-level security requirements. Users can verify with a multi-step flow, and teams can enforce 2FA for organization members.

We also improved related settings UI and made the verification flow more resilient with better token handling and time drift tolerance.

Billing and subscription improvements

Billing became more transparent and complete. Updates included invoice display, invoice webhook handling, pending downgrade banners, promotion code support, pay-as-you-go language, and clearer pricing/plan limits.

These changes make it easier for teams to understand current usage, billing status, and what happens when a plan changes.

CLI webhook support

We added CLI connection and pending channel support, plus localhost webhook flows in the console. This helps developers test event-driven workflows locally without needing to deploy a public webhook endpoint for every iteration.

Request logs and callback delivery views were also improved so users can see how webhook and WebSocket events move through the system.

API key and webhook lifecycle controls

API keys and webhooks gained safer lifecycle behavior, including soft deletion/revocation support and API key expiration. This gives teams better control over credentials and integrations without losing useful history immediately.

Request log detail pages

We added a request log detail page with failover timelines, preview tabs, and metrics. This gives users a clearer view of what happened during an AI request, especially when a request includes provider attempts, callbacks, or WebSocket delivery.

Rate limiting and provider metadata

We implemented a unified rate limiting system and updated launch-ready plan limits. We also refreshed provider pages, model/provider admin views, and structured output compatibility for Anthropic.