Exciting news! @Alibaba_Qwen Qwen3.8-Flash-Next running at 100% on a 96GB @Apple Mac Studio M3 Ultra (including the ANE).
The model uses a custom Trellis quantization, retaining nearly original quality. Optimized for real-world workloads and concurrency, enabling simultaneous prefill + decode (prefill on ANE, decode on GPU) while maintaining 60+ decode TPS during prefill.
Incredible to see this machine pushed to its absolute limits with such an amazing model. More updates soon!
Thanks @QwenDevs for opening this beautiful model to the world, eagerly waiting for Qwen4! 🤩
Who said you can't run a top-tier 27B reasoning model on a $150 GPU from 2016? 🤯
It’s live: run OrcaSAQ-2-27B on an NVIDIA Tesla P100 16GB via a plug-and-play Docker image!
Huge thanks to @OrcaRouter for releasing such an incredible model.
Powered by the footless engine + custom Pascal kernels:
• ~30 tok/s decode w/ MTP speculative decoding (~20 tps standard)
• ~200 tok/s prefill | up to ~100k context
• OpenAI-compatible API + JEV-like decision endpoint
• Single-stream for now
It’s experimental! But if this gets some love, I'm going to keep pushing on multi-P100 and other GPUs like the V100 and maybe the A100.
Repo & setup: https://t.co/1tbveQXeDf
I'm going to drop some sample generations in the replies 👇
Our #SteamNextFest playtest runs Oct 14-21, and we just hit 2.1M wishlists 🥳🍾
To say thanks, we’re giving away 5 DF hats, 5 DF bags, and
🚨1 #NVIDIA RTX 4060ti🚨
To win:
1⃣Follow us
2⃣Like, RT & comment on this post
3⃣Random winners picked Oct 21
#1WeekWarriors#deltaforce
Ya llegó el sorteo de febrero 💥
💸 Ganar 100€ para comprarte lo que quieras en Eneba
🏆 Ingresa aquí para participar: https://t.co/L97doewUgv
✍️ Escribe qué es lo que te comprarías: juegos, suscripciones, tarjetas prepago y más
🌍 Válido para España y Latinoamérica
Muchas gracias @3dlac y a @Control_3D por sortear una PRUSA i3 MK3S+!
Tengo más ilusión que con el sorteo de Navidad 😱 #3dprinter
https://t.co/goStlSwXtq