Kev on AMD ROCm: 171 ms decision model on a Radeon PRO W7900
We ran the open-weights Kev decision models on an AMD Radeon PRO W7900 under Windows. Accuracy matched the published model cards to…
Tag
Tag: local AI
NVMe vs SATA LLM model loading: 10 runs per drive, measured
A 19.92 GiB Qwen model loaded 2.25x faster from NVMe than SATA, yet time to first token was identical. Cold-start benchmark with…
Clone Windows to an NVMe SSD: a verified SATA-to-NVMe migration
A full SATA-to-NVMe migration on an HP Z6 G4 workstation, with byte-level medium verification, Rescuezilla disk-to-disk cloning and measured before-and-after throughput.
SSD storage for local AI: how much you need and what to buy in 2026
Local models eat disk space faster than most people plan for. Measured throughput, real model sizes and an August 2026 price snapshot…
LM Studio CLI: full automation — downloading, configuring and running Qwen models without GUI
Complete guide to LM Studio CLI on Windows 11. Commands lms get, load, unload, rm — automated Qwen model management without clicking…
Free local AI video on AMD: text-to-video with Wan 2.1 in ComfyUI
How to use the Google Veo 3 API via Python to generate AI video. Costs, limitations, generation time and a hybrid approach…
LM Studio as a local backend for an agent: API, code editors and developer scenarios
How to use LM Studio as an OpenAI-compatible API backend for coding agents. Integration with Continue.dev, Cursor and a custom Python script.…
Multi-Token Prediction on AMD W7900: 1.85× speedup with llama-server Vulkan (Qwen3.6-35B benchmark)
We benchmark Multi-Token Prediction in llama-server (llama.cpp, Vulkan) on AMD PRO W7900. Real results: 49.7 → 91.8 tok/s, up to 2× on…
Private AI Assistant for Productivity: How Freelancers and Managers Can Get Work Done in the Background
A practical English article on private AI assistants, background work, personal context, privacy, and real productivity gains for freelancers and managers.