My RTX 4090 just hit 181 tok/s on Qwen 3.8 27B model. O Monday I thought 140 was the…
This is a AI post classified by Jev as Open & local models (a model release), kept by the AI Radar because it carries real work, not commentary.
My RTX 4090 just hit 181 tok/s on Qwen 3.8 27B model. O Monday I thought 140 was the ceiling. I was wrong. @mr_r0b0t built a different engine. I wanted to push the card again to see if I can squeeze more out of it. The headline number moved 30%, the like-for-like number moved 4%, and the gap between them is the actual picture. Here is my full breakdown of whether the recipe has legs. A DIFFERENT ENGINE Two pieces. A better quant and a better drafter. The quant is EXL3 at 4.0 bits per weight, 15.4 GiB for the whole 27B model. The drafter is DFlash2, a block-diffusion model that proposes u
Posted by Yume_X (2.1k followers) 2 days ago · 80 likes · 5.9k views · view the original post on X. Kept by the AI Radar as Open & local models.
More AI work like this
- dear rtx 3060 owners, and every 12gb card behind it. bonsai 2 27b dense went from 26 to… — @sudoingX
- Nimble is able to process images! — @madiator
- Woah...it claims to be open Jev. And it's 100% opensource. — @Saboo_Shubham_
- your rtx 3060 was running bonsai 2 at 26 tok/s this morning and now does 40 tok/s, i… — @sudoingX
- Meet DeepSeek-V4.1-Flash: a blazing fast image-text-to-text model. It takes both images… — @HuggingModels
- Meet Edge0-35B-A3B-preview: a 35B parameter MoE model built for edge inference. It runs… — @HuggingModels
- Putting Jeff-1 out into the world, this is an attempt at a Jev like model, built on qwen… — @DJLougen
- Train a 9M parameter language model from scratch in five minutes. — @tom_doerr
Every post is read and classified by Jev (TypeSafe): what it is, which market it belongs to, and whether the link is a real tool. 27.2k posts from 4.8k X accounts over the last 14 days, 1.2k tools, 19 markets. Collected every 5 minutes, fully re-ranked every hour — last update 2026-09-19 18:08 UTC. Full method.