Sharing notes on various things I expect to happen in the rest of 2026:

1. Open models haven’t had their true agent moment like Opus 4.5

2. Gemini still doesn’t have a meaningful competitor for Claude Code and Codex

3. I don’t expect an open-weights Mythos this year

4. American open models are slowly gaining steam

5. Anthropic and OpenAI are just getting up to speed in model iterations

6. More existing power structures will assert themselves on AI

+my thoughts on the coming social disruption of AI in the US.