Building the fastest LLMs: why we're starting with diffusion

(celeris.ai) by peter_d_sherman | Sep 19, 2026 | 0 comments on HN