Ollama CEO says the gap between open and frontier AI models has shrunk to about four months
In his first interview in Korea, Jeffrey Morgan told etnews the gap was 14 months a year ago and that 9 million developers now use Ollama.
Why it matters
Ollama is a popular tool for running open-weight AI models on personal computers and private servers. In his first interview in Korea, CEO Jeffrey Morgan told etnews the gap between open and Big Tech frontier models has narrowed from 14 months a year ago to about four, with some open models beating top models on specific benchmarks. He expects open models to handle 80–90% of AI tokens, leaving the hardest work to closed models from OpenAI, Anthropic and Google. Ollama held its first event outside North America in Seoul, partnered with FriendliAI and plans local hosting in Korea.
Key facts
- Open–frontier gap: 14 months a year ago, about 4 now
- 9 million developers use Ollama
- Open models could handle 80–90% of AI tokens
- First Ollama event outside North America, about 700 attendees
- Partnership with Korean inference firm FriendliAI
- Local hosting in Korea planned