AI Radar
Support
LiveUpdated 2026-09-19 18:08 UTC

GLM5.3-flash got 44% faster thanks to @MiaAI_lab and @plotarmordev

GLM5.3-flash got 44% faster thanks to @MiaAI_lab and @plotarmordev New day, new GLM-5.3 Flash EXL3 recipe from Mia —…

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.

GLM5.3-flash got 44% faster thanks to @MiaAI_lab and @plotarmordev New day, new GLM-5.3 Flash EXL3 recipe from Mia — so it went through the same gauntlet as every release: idle server, temp 0, her own bench script, same-day stock A/B on my 2× DGX Sparks. Cooperative decode MoE (geo-1): Prose ×1: 31.7 → 35.4 (+12%, within noise of Mia's 36–37) Prose ×2: 39.8 → 57.3 (+44%, above her mark) Prose ×4: 69.3 → 80.4 (+16%, above her 75–67) Structured ×1: 74.6 → 79.9 (+7%) And from yesterday's drop, now on this kit: weight load 346s → 48s. One tip for other kit owners: rebuilds of cooperative_moe

Posted by JCode (3.6k followers) 1 days ago · 2 likes · 188 views · view the original post on X. Kept by the AI Radar as Open & local models. Tools mentioned: glm-5.3-flash-exl3-2x-dgx-sparks.

More AI work like this

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.