Learn about dLLMs - Fireside Chat with Stefano Ermon (Founder & CEO, Inception Labs) and Irving Hsu (Investor, Mayfield)

​What if AI didn’t generate text one word at a time but could generate much more tokens in parallel? Diffusion transformed image generation - now it may transform language.

​Join us for a deep dive into diffusion large language models (dLLMs) with Stanford Professor Stefano Ermon, founder and CEO of Inception Labs and Irving Hsu from Mayfield, an early investor in Inception.

​Stefano co-invented diffusion models and he’s applying that breakthrough to text with Inception’s Mercury models, which generate responses in parallel rather than sequentially, unlocking dramatically faster and more cost-efficient AI systems.

​The company launched Mercury 2 this week - the fastest reasoning LLM and first reasoning dLLM. It delivers 1000 tokens per second, 5x faster than leading speed-optimized reasoning models.

​Inception raised $50 million in seed funding, led by Menlo Ventures, with participation from Mayfield, Innovation Endeavors, M12, Snowflake Ventures, Databricks, and NVIDIA’s venture arm.