@no_stp_on_snek: this week just keeps getting better.
Summary
Molei Tao introduces FLARE, a diffusion language model that achieves near GPT5 performance with significantly faster inference speed.
View Cached Full Text
Cached at: 08/14/26, 05:52 PM
this week just keeps getting better.
Molei Tao (@MoleiTaoMath): Introducing FLARE, a diffusion language model, with near GPT5 performance, but order of magnitude faster -
Those in this domain may know
- LLaDA & Dream: pioneer, used by many other research
- Mercury: 1st commercial model, try it to believe!
- Nemotron Diffusion,
Similar Articles
@ying11231: Impressive performance on TPU.
A blog post from LMSYS Org details optimizing Ling-2.6-1T, a 1 trillion parameter hybrid MoE model, on TPU v7x using SGLang-JAX, achieving efficient inference by hiding MoE data movement behind computation with a single Pallas kernel.
@SebastienBubeck: https://x.com/SebastienBubeck/status/2075596982622835006
GPT-5.6 significantly outperforms published state-of-the-art on a fundamental mathematical problem about gradient flow length, achieving exponential improvements. This marks a major advance in AI's ability to reason about complex mathematical questions.
Introducing celeris-1 (2 minute read)
Celeris-1 is a new language model using diffusion-based inference architecture, achieving near-GPT-5 level intelligence with 15x faster response times and high throughput.
Introducing GPT-5.1 for developers
OpenAI releases GPT-5.1, a new model in the GPT-5 series that dynamically adapts thinking time based on task complexity, offering 2-3x faster performance than GPT-5 while maintaining frontier intelligence. The release includes extended prompt caching (24-hour retention), new coding tools (apply_patch and shell), and a 'no reasoning' mode for latency-sensitive applications.
@philipkiely: https://x.com/philipkiely/status/2069212319746506968
Baseten announces the world's fastest API for the GLM-5.2 open model, achieving over 280 tokens per second via NVFP4 quantization, disaggregated inference, and other optimizations.