ZDNET Writer Switches to Ollama, Citing 16GB Local AI Privacy Over ChatGPT

3 articles · Updated · ZDNet · May 26

ZDNET’s latest report says its author has stopped using ChatGPT and now relies on Ollama, a locally installed AI tool positioned as free, private and open-source.
Ollama runs models directly on Windows, macOS or Linux hardware, with the report recommending 16GB RAM and faster performance from Nvidia GPUs with 8GB-plus VRAM or Apple Silicon Macs.
The author argues local use avoids cloud data collection, works offline and can even run on an air-gapped machine, making it more suitable for privacy-sensitive research.
Ollama’s appeal also rests on flexibility: users can download multiple models—including DeepSeek, Gemma, Qwen, Mistral and Llama—or host one instance on a LAN server for several devices.
The piece frames that local-first setup as an alternative to cloud AI’s cost and infrastructure burden, citing a report that data centers now consume 6% of U.S. electricity.

Open-source AI now rivals GPT-4 on laptops. Is the era of big tech's cloud AI dominance already coming to an end?

Reports show AI coders fail more often. Are local LLMs a productivity revolution or just a faster way to create broken code?

Local AI promises total privacy, but can it be trusted after the 'Bleeding Llama' flaw exposed thousands of users?