What Inception does
Inception is an AI research and product company building diffusion-based language models (dLLMs) aimed at reducing the latency and cost bottlenecks of autoregressive LLMs. In Inception’s framing, standard “frontier” LLMs generate tokens sequentially left-to-right, which makes speed and cost scale with the number of tokens and steps required by multi-step agent workflows.
More
In contrast, diffusion generates through a coarse-to-fine refinement process, enabling parallel refinement and earlier revisions during generation. Inception’s flagship products are part of the Mercury model family (including Mercury Coder for coding and Mercury 2 for reasoning/chat use cases), delivered through an OpenAI-compatible API and available for enterprise deployment. Inception positions Mercury as a drop-in replacement for existing LLM workflows—supporting RAG, tool use, and agentic loops—while targeting throughput and latency requirements for production systems such as interactive coding/editing, agent pipelines, real-time voice agents, and search/RAG workflows. The company’s go-to-market is primarily developer and enterprise access via its API and enterprise-ready deployments, with partnerships and customer integrations highlighted in its blog (for example, SearchBlox powering real-time GenAI search using Mercury).