Author

Tech-Practice

@ttio2tech_28094

articles: 12
total claps: 142
top claps: 56
avg read: 3.3 min
paywalled: 12
active: 2025-02-12 → 2026-07-17

Wiki topics

OPS · LLMOps & Inference · 7 LLM · Large Language Models · 6 AI · AI · General · 4 MM · Multimodal & Generative Media · 4 🔓 · Open Source · 4 🏃 · Running & Endurance · 2 💻 · Programming · 2 EVAL · Evaluation & Benchmarks · 1

Top tags

#5090 ·8 #llm ·6 #gpu ·6 #flux ·3 #ai ·3 #vllm ·2 #nvidia ·2 #rtx-3090 ·2 #blackwell ·1 #unsloth ·1

AI SPT

TE Tech-Practice

RTX 5090's speed for GenAI — Flux1.dev,

How fast is the RTX 5090 running text to image or video?

Feb 26, 2025 · 3 min read · 56 claps paywalled #genai #gpu #5090

TE Tech-Practice

What makes the Nvidia 5090/5080 unique?

Nvidia just released their new 5000 GPUs. How are they different from previous generation cards?

Feb 12, 2025 · 2 min read · 51 claps paywalled #nvidia #5090 #gpu

AI TCH

TE Tech-Practice

How to Install RTX 5090 / 5080 /5070 Ti / 5070 drivers on Linux — detailed guide

For best performance, you will need to run the 5090 / 5080 / 5070 Ti / 5070 or other 50 series GPU on Linux (or WSL on Windows). This…

Feb 18, 2025 · 3 min read · 13 claps paywalled #rtx #rtx-5090 #5090

How to Install RTX 5090 / 5080 /5070 Ti / 5070 drivers on Linux — detailed guide

AI ART

TE Tech-Practice

Flux.1-Kontext — image editing made easy

New model from the Flux model maker — Black Forests labs! Have fun playing with it, but also be cautions that it can make mistakes.

Jun 06, 2025 · 3 min read · 13 claps paywalled #flux #image-editing #black-forest-labs

Flux.1-Kontext — image editing made easy

AI TCH

TE Tech-Practice

Consumer GPU run Flux.1-Kontext — open source image editing model

Installation and the performance.

Jun 26, 2025 · 2 min read · 3 claps paywalled #kontext #flux #comfy-ui

Consumer GPU run Flux.1-Kontext — open source image editing model

TE Tech-Practice

I Ran a 428-Billion Parameter AI Model on My Desk — Here’s What It Took

Building a 4x RTX 3090 rig to run MiniMax-M3, one of the most capable open-weight models ever released, fully offline.

Jul 17, 2026 · 6 min read · 2 claps paywalled #llm #rtx-3090 #minimax

I Ran a 428-Billion Parameter AI Model on My Desk — Here’s What It Took

SPT AI TCH

TE Tech-Practice

I found the best DFlash Setting on Gemma 26B — 228 to 600 tok/s on One RTX 5090

Speculative decoding is one of the most effective inference optimization techniques available right now. The idea is elegant: a small draft…

May 08, 2026 · 5 min read · 1 claps paywalled #dflash #llm #5090

I found the best DFlash Setting on Gemma 26B — 228 to 600 tok/s on One RTX 5090

AI TCH

TE Tech-Practice

Free LLM (Deepseek, Kimi-K2-Thinking, Qwen3, GLM-4.6) via iFlow CLI coding agent

Steps to use these top open source LLM completely free!

Nov 18, 2025 · 3 min read · 1 claps paywalled #iflow #glm #llm

Free LLM (Deepseek, Kimi-K2-Thinking, Qwen3, GLM-4.6) via iFlow CLI coding agent

AI SPT

TE Tech-Practice

VLLM runningGLM-4.7-Flash: How To Run Locally (step by step guide)

The Thrill of Running Cutting-Edge LLMs Locally

Jan 20, 2026 · 3 min read · 1 claps paywalled #5090 #vllm #llm

VLLM runningGLM-4.7-Flash: How To Run Locally (step by step guide)

MDA AI TCH

TE Tech-Practice

Turns out Flux.2 is fake open source — How Black Forest Labs is Gaslighting the AI Community

They’re using the language of open innovation to build a walled garden. Don’t fall for it.

Nov 26, 2025 · 3 min read · 1 claps paywalled #flux #aiimages #gpu

Turns out Flux.2 is fake open source — How Black Forest Labs is Gaslighting the AI Community

TE Tech-Practice

I Benchmarked Qwen3.6–35B-A3B Model on 3090, 4090, 5090 and M5 Max. Here’s What Nobody Tells You.

RTX 3090, 4090, 5090, and Apple Silicon (M5 Max) — same model, same engine, brutally honest results.

Apr 21, 2026 · 5 min read · 0 claps paywalled #rtx-3090 #5090 #m5-max

I Benchmarked Qwen3.6–35B-A3B Model on 3090, 4090, 5090 and M5 Max. Here’s What Nobody Tells You.

TE Tech-Practice

Easy finetuning LLM on Blackwell — unsloth is now supported on Blackwell (RTX 5070, 5080, 5090)

Glad to see that unsloth is able to run on the Blackwell hardware.

Jul 04, 2025 · 2 min read · 0 claps paywalled #blackwell #5090 #unsloth