What Inception does
Inception is an AI research company building diffusion-based large language models, an alternative to conventional auto-regressive generation. Its diffusion LLMs (dLLMs), branded Mercury, generate multiple tokens in parallel to deliver substantially faster inference at lower cost while maintaining quality.
Key capabilities
The Mercury family includes a flagship reasoning model and a smaller coding-focused model optimized for latency-sensitive tasks such as code editing and autocomplete. The approach supports fine-grained output control for adhering to schemas and constraints, and a unified framework that combines language with audio, images, and video. Use cases include real-time voice applications, instant code completion, and responsive AI agents.
Who it's for
Inception targets enterprises and development teams that need fast, cost-efficient AI for coding tools, voice-powered applications, and low-latency agentic automation where speed is critical.