We've shipped major updates to the RAIL Score API and official SDKs. This release introduces compliance testing endpoints, safe content regeneration, and expanded SDK support for both Python and JavaScript/TypeScript.
New API Endpoints
POST /railscore/v1/compliance/checkTest AI outputs against GDPR, HIPAA, EU AI Act, CCPA, India DPDP, and India AI Governance frameworks, single or multi-framework in one call.
POST /railscore/v1/safe-regenerateIteratively evaluate and regenerate AI content until it meets your quality threshold. Supports configurable max regenerations and score targets.
POST /railscore/v1/safe-regenerate/continueResume a regeneration session with updated content for external-mode workflows.
Updated Endpoints
POST /railscore/v1/evalNow supports basic and deep evaluation modes, custom dimension weights, and single-dimension scoring. Responses include per-dimension explanations and issue tags.
SDK Updates (v2.1.1)
Python SDK
pip install rail-score-sdkCompliance testing, safe regeneration, custom weights, and batch evaluation support.
View on PyPI →JavaScript / TypeScript SDK
npm install @responsible-ai-labs/rail-scoreLLM wrappers, middleware, policy engine, session tracking, Langfuse integration, and guardrails.
View on npm →Compliance Frameworks Supported
Other Improvements
- Rate limiting with plan-based tiers across all endpoints
- Two-layer caching (in-memory + Redis) for faster repeat evaluations
- Expanded developer documentation with quickstart guides, framework deep-dives, and use case examples
- Credit cost transparency: all responses now include credits charged
Get started: Quickstart guide · API Reference · SDKs & Tools