Tweet by thomas_ankcorn

May 2, 2026

The local llm folks need to understand what this means. Shared kv caches make the economic reality of hosting your own model stupid. Just worked out the utilisation required to beat Cloudflares pricing for workers ai 😬 https://t.co/iXPb7RI9Ss

Author
thomas_ankcorn
Date
May 2, 2026