Do Diffusion LLMs Scale? Testing Mercury 3 + Decide

Why watch

Technical deep-dive testing Inception Labs' Mercury 3 diffusion LLM and Mercury Decide decision model, comparing their 1,068 tokens/sec performance against benchmarks. Covers independent evaluations, real-time speed tests including 5,000-word essays and agentic tool-calling, and demonstrates the emerging pattern of fast diffusion models paired with deliberate decision models. For AI researchers and engineers evaluating next-generation LLM architectures.