Some ideas for what comes next, May 2026
The AI landscape in May 2026 is expected to see continued rapid advancement and increasing real-world consequences, with no breaks anticipated from the current trajectory of disruption and surprise. A key area of focus is the performance gap between closed and open-weight models, particularly in agentic capabilities, with the author predicting that open models may take longer than expected to reach parity with leading closed models like Anthropic's Opus 4.5. Google's Gemini 3.5 Flash is noted as not yet a direct competitor to specialized tools like Claude Code and Codex, suggesting that while Gemini may serve Google's internal products, open models might primarily find use in automated enterprise agents rather than as primary tools for knowledge work. AI
IMPACT Author predicts open models will take longer to match closed models, potentially shifting their use cases to enterprise agents.