Celeris-1: A diffusion LLM benchmarked at 2,082 output tokens/s

(artificialanalysis.ai) by jn2clark | Aug 5, 2026 | 0 comments on HN