Critique of Ollama's Practices in Local LLM Ecosystem

Founded in 2021 by Jeffrey Morgan and Michael Chiang, Ollama gained traction as an accessible interface for local large language models. Its inference capabilities were based on llama.cpp, developed by Georgi Gerganov in March 2023. For more than a year, Ollama's documentation failed to reference llama.cpp or include MIT license notices. GitHub issues, such as issue #3185, remained unresolved for over 400 days. In mid-2025, Ollama transitioned to a custom ggml implementation, which introduced various bugs. Community benchmarks indicated that llama.cpp performed 1.8 times faster. The company faced backlash for deceptive model naming and launched a closed-source GUI app in July 2025 without a license. Late 2025 saw the introduction of cloud-hosted models, which raised privacy issues, including CVE-2025-51471. Alternatives like llama.cpp provide similar features, while Gerganov's ggml.ai joined Hugging Face in February 2026.