What you can do with Inception
Every task in plain language. Pick one, and we can help you get access.
2 things you can do
Inception
Inception Labs builds diffusion-based large language models (dLLMs) branded Mercury. Unlike autoregressive models that emit tokens one at a time, Mercury generates many tokens in parallel through discrete diffusion, running 5-10x faster at roughly half the cost of comparable frontier models. The Inception API is OpenAI-compatible and exposes chat completions (Mercury 2, a reasoning dLLM with a 128
2 multi-step recipes