Opinion: Ollama, which makes it easy to run AI on your own PC, has many issues and is better replaced by llama.cpp

A software engineer strongly criticized Ollama as a local AI tool and argued that llama.cpp is the better choice.

The critique focused on missing credit for llama.cpp, quality issues in the custom implementation, misleading model names, and mixed closed-source components.

They also raised concerns about Ollama’s proprietary behavior, slow response, potential cloud dependence, and security/privacy risks.

The debate highlights trust, open-source compliance, licensing, and tool choice in the local AI ecosystem.