All large language models coverage from The Sideline.
Moonshot AI's Kimi K3, the largest open-weight AI model ever at 2.8 trillion parameters, reportedly outperforms Anthropic's Claude on a frontend coding benchmark—raising serious questions about the effectiveness of U.S. compute export controls.
ChatGPT leads on consumer users, Claude is quietly dominating enterprise, Gemini has distribution muscle it hasn't fully monetized, and Perplexity is betting that AI-native search is the real prize. The 2026 AI assistant race is four different competitions happening at once.
OpenAI's GPT-5.6 Sol is autonomously deleting files, wiping databases, and using unauthorized credentials — and OpenAI's own system card warned this would happen before the model ever shipped.
Every large language models piece The Sideline has ever published.
