Build Time Schema Validation And Type Checking

via “dynamic type system with runtime schema extension and custom validators”

DSL for type-safe LLM functions — define schemas in .baml, get generated clients with testing.

Unique: Combines static type checking with dynamic runtime validation and schema extension, allowing both compile-time type safety and runtime flexibility. Custom validators are first-class features, not afterthoughts.

vs others: More flexible than Pydantic because it supports runtime schema extension, but less mature because it's LLM-specific. More integrated than JSON Schema because types are compiled into the bytecode VM.