Diffusion-Models on Onur Baydaşhttps://onurbaydas.com/en/tags/diffusion-models/Recent content in Diffusion-Models on Onur BaydaşHugoen-usSun, 18 Oct 2026 14:58:27 +0300High-speed diffusion LLM performance on a single RTX 5090 with ORLhttps://onurbaydas.com/en/news/2026-10-18-orl-diffusion-llm-performance/Sun, 18 Oct 2026 14:58:27 +0300https://onurbaydas.com/en/news/2026-10-18-orl-diffusion-llm-performance/New research called ORL aims to significantly speed up diffusion language models (dLLMs) without sacrificing generation quality. This work demonstrates high performance, achieving 2,788 tokens per second on a single RTX 5090 GPU.