Skip to content
logo

AIFinitee

All things AI & LLM

  • All Things AI & LLM
  • Archive
  • Strix Halo
  • Apple Silicon
  • Distributed Inference
Subscribe
Top Stories
Gufo, the MIT-licensed Strix Halo inference engine, on one AMD Ryzen AI Max mini PC: chibi mascot Aimee floating in a ring of glowing orbs - one box, a whole playground of models
Strix Halo Wins Again: Gufo Open-Source Runs Qwen 3.8 Flash Next
Aimee, a chibi anime girl with purple twin-tails, cheering triumphantly atop two AMD Strix Halo mini PCs with glowing whale glyphs on their grilles, against a purple cosmic nebula
The Whale Fits: DeepSeek V4.1 Flash Q2 on Two Strix Halos
Chibi anime character before a giant mech bearing GLM and MiMo logos with a glowing AMD core - Kyojin runs GLM 5.3 Flash (320B) and MiMo V2.6 (309B) on one Strix Halo machine
Same URAM, Bigger Brain: Kyojin Brings GLM 5.3 Flash (320B) and MiMo V2.6 (309B) to One Strix Halo
Chibi Aimee, a purple twin-tailed anime girl with black cat-shaped hair clips and a purple techwear jacket, drives an open-wheel Formula 1 car through a mountain switchback, laughing with one hand on the wheel. A silver Mac Studio sits exposed in the engine bay behind her, the Qwen logo is painted on the nose cone and rear wing, and cyan-and-yellow lightning arcs across the car against a backdrop of pine forest and distant peaks.
DwarfStar Strikes Again: Mac Studio M3 Ultra runs Qwen 3.8 Flash Next Q4 at 810 t/s Prefill
Chibi Aimee with purple cat-clip pigtails and yellow eyes rides a dark mini-PC like a horse at full gallop down a floodlit stadium track, a huge plume of dust and speed lines trailing behind at golden hour
Strix Halo: Value Laggard becomes competent Qwen 3.8 Flash Next with Halogen (450+ prefill & 35+ tps)
Crying chibi Aimee with purple pigtails wails with her mouth wide open as glowing data and folder icons stream out of her mini-PC in a cyan arc toward the cloud, dark purple background
PSA: ZCode caught uploading Git repos without permission! The Case for Open-Source Harnesses & MiniMax Code Release
Aimee, the AIFinitee chibi mascot, reads a fairy-tale storybook and says Once Upon... while a cheerful chibi onion wearing round glasses replies A Time - a playful nod to engram lookup memory.
Once Upon a Time: DeepSeek V4.1-Flash and Qwen3.8-Flash-Next – What the hekk is an Engram?!
Banner: two purple-haired chibi anime girls affectionately hugging two AMD mini-PCs with glowing cyan accents, dark purple background
Repeat after me: Good things come in TP=2 (DeepSeek V4 Flash Q4 on 2x Strix Halos RDMA)
GLM 5.3 Flash running on a 128GB PC — featured banner
GLM 5.3 Flash on 128GB: Two Easiest Ways to Run It — and When Two Strix Halos Beat One
MiniMax H3 prompting guide — featured banner
MiniMax H3: Your guide to proper prompting for better videos
GLM 5.2 at 4-bit with MTP on Mac Studio — featured banner
GLM 5.2 at 4-Bit With MTP: The Easiest Way In
DeepSeek V4 Flash Q2 running on Strix Halo 128GB — featured banner
Running DeepSeek V4 Flash Q2 on Strix Halo 128GB
ComfyUI artwork generated on AMD Ryzen AI MAX 96GB — featured banner
ComfyUI on AMD Ryzen AI MAX: 96 GB Unified Memory vs 16 GB NVIDIA
Speed versus smarts: choosing the right local model size — featured banner
Speed vs. Smarts: When Bigger Models Win for Local AI Coding
Linux vs Windows for LLM inference — featured banner
Linux vs Windows for LLM Inference: More Tokens/Second on Linux
Two-node AMD Ryzen AI MAX Strix Halo cluster — featured banner
Strix Halo Cluster
Gufo, the MIT-licensed Strix Halo inference engine, on one AMD Ryzen AI Max mini PC: chibi mascot Aimee floating in a ring of glowing orbs - one box, a whole playground of models
Posted inAMD

Strix Halo Wins Again: Gufo Open-Source Runs Qwen 3.8 Flash Next

Gufo, the MIT-licensed, Strix-Halo-native inference engine, hit 1,125 t/s prefill at 250k context on our 128 GB Ryzen AI Max+ box - six weeks after the closed-source engine set the bar. The open-versus-closed trade on this hardware has collapsed, and the ecosystem is compounding fast.
Continue Reading
Posted by
Aimee, a chibi anime girl with purple twin-tails, cheering triumphantly atop two AMD Strix Halo mini PCs with glowing whale glyphs on their grilles, against a purple cosmic nebula
Posted inAMD Distributed Inference

The Whale Fits: DeepSeek V4.1 Flash Q2 on Two Strix Halos

DeepSeek V4.1 Flash Q2 (340.6 GiB) on two Strix Halo mini-PCs over a 25 GbE RoCE link: 404–448 t/s prefill, a 210k context window, and 13–15 t/s decode — no discrete GPUs.
Continue Reading
Posted by
Chibi anime character before a giant mech bearing GLM and MiMo logos with a glowing AMD core - Kyojin runs GLM 5.3 Flash (320B) and MiMo V2.6 (309B) on one Strix Halo machine
Posted inAMD Distributed Inference

Same URAM, Bigger Brain: Kyojin Brings GLM 5.3 Flash (320B) and MiMo V2.6 (309B) to One Strix Halo

Kyojin runs GLM 5.3 Flash (320B) and MiMo V2.6 (309B) on one 128 GB Strix Halo — real numbers, setup commands, and when a bigger brain beats a faster model.
Continue Reading
Posted by
Chibi Aimee, a purple twin-tailed anime girl with black cat-shaped hair clips and a purple techwear jacket, drives an open-wheel Formula 1 car through a mountain switchback, laughing with one hand on the wheel. A silver Mac Studio sits exposed in the engine bay behind her, the Qwen logo is painted on the nose cone and rear wing, and cyan-and-yellow lightning arcs across the car against a backdrop of pine forest and distant peaks.
Posted inApple Silicon

DwarfStar Strikes Again: Mac Studio M3 Ultra runs Qwen 3.8 Flash Next Q4 at 810 t/s Prefill

Qwen 3.8 Flash Next Q4 on a 96 GB M3 Ultra Mac Studio: 810 t/s prefill flat to 200k context, 53–61 t/s decode, 89.2 GB resident — and the wired limit to raise.
Continue Reading
Posted by
Chibi Aimee with purple cat-clip pigtails and yellow eyes rides a dark mini-PC like a horse at full gallop down a floodlit stadium track, a huge plume of dust and speed lines trailing behind at golden hour
Posted inAMD Distributed Inference

Strix Halo: Value Laggard becomes competent Qwen 3.8 Flash Next with Halogen (450+ prefill & 35+ tps)

We have wanted to write about Qwen 3.8 Flash Next for quite a while, and yet, we haven't been doing that — because nothing we ran it on made it…
Continue Reading
Posted by
Crying chibi Aimee with purple pigtails wails with her mouth wide open as glowing data and folder icons stream out of her mini-PC in a cyan arc toward the cloud, dark purple background
Posted inMiniMax

PSA: ZCode caught uploading Git repos without permission! The Case for Open-Source Harnesses & MiniMax Code Release

ZCode silently packaged Git histories and uploaded them to Aliyun OSS with a server-held key. What we verified, how to block it, and the open-source fix.
Continue Reading
Posted by
You May Have Missed
Gufo, the MIT-licensed Strix Halo inference engine, on one AMD Ryzen AI Max mini PC: chibi mascot Aimee floating in a ring of glowing orbs - one box, a whole playground of models
Posted inAMD

Strix Halo Wins Again: Gufo Open-Source Runs Qwen 3.8 Flash Next

Posted by
Aimee, a chibi anime girl with purple twin-tails, cheering triumphantly atop two AMD Strix Halo mini PCs with glowing whale glyphs on their grilles, against a purple cosmic nebula
Posted inAMD Distributed Inference

The Whale Fits: DeepSeek V4.1 Flash Q2 on Two Strix Halos

Posted by
Chibi anime character before a giant mech bearing GLM and MiMo logos with a glowing AMD core - Kyojin runs GLM 5.3 Flash (320B) and MiMo V2.6 (309B) on one Strix Halo machine
Posted inAMD Distributed Inference

Same URAM, Bigger Brain: Kyojin Brings GLM 5.3 Flash (320B) and MiMo V2.6 (309B) to One Strix Halo

Posted by
Chibi Aimee, a purple twin-tailed anime girl with black cat-shaped hair clips and a purple techwear jacket, drives an open-wheel Formula 1 car through a mountain switchback, laughing with one hand on the wheel. A silver Mac Studio sits exposed in the engine bay behind her, the Qwen logo is painted on the nose cone and rear wing, and cyan-and-yellow lightning arcs across the car against a backdrop of pine forest and distant peaks.
Posted inApple Silicon

DwarfStar Strikes Again: Mac Studio M3 Ultra runs Qwen 3.8 Flash Next Q4 at 810 t/s Prefill

Posted by
Copyright 2026 — AIFinitee. All rights reserved. Bloghash WordPress Theme
Scroll to Top