Slam-1 is a Speech Language Model that combines LLM architecture with ASR encoders for superior speech-to-text transcription. It is optimized for customization via prompting and delivers enhanced accuracy by understanding context and semantic meaning. It is currently in public beta.