# gemini

Google AI's weekly roundup covered Maps, Workspace, Gemini Embedding 2, Gemini API controls, Chrome rollout, and a breast cancer research collaboration with Imperial College London and the UK NHS. One tweet spans product, dev tools, and research.

Simon Willison shared Gemini API's new spend caps feature, saying it's great news for anyone running Gemini prompts in CI or letting agents experiment with the API — less fear of surprise bills.

OpenClaw's ACP lets you spawn Codex, Claude Code, and Gemini from Discord/Telegram chat. Now with Telegram topic binding, persistent bindings that survive restarts, ACP Provenance for audit trails, and more. (Updated 2026-03-09)

Google announced Gemini 3.1 Pro preview with stronger core reasoning and a verified 77.1% on ARC-AGI-2. The real engineering question is whether the model can handle complex multi-step production workflows, not just win benchmarks.

Ethan Mollick's game-changing AI framework: Model, App, Harness. The same AI (e.g., Claude Opus 4.6) performs vastly differently across layers. Mollick used Claude Code to turn GPT-1's 117M weights into 80 books in ~1 hour, selling out immediately.

SWE-bench: Claude Opus 4.5 (76.8%) unexpectedly beat 4.6 (75.6%) for #1. MiniMax M2.5 tied for #2 at 1/20th Opus's price, with 4 Chinese models in top 10. GPT-5.3-Codex missed due to no API. Bonus: Claude for Chrome to add chart labels.