Google Gemini & Workspace — TurboQuant/Agent Designer/Pentagon/Apple Siri iOS 27, 750M MAU, Mac/3.1 Flash Live/Vids free Veo tool + Gemma 4 open agentic models (Ollama/MLX quants + Android/iOS local) + Agent mode OpenClaw alt + Projects folders
Key Questions
What new organizational features does Google Gemini offer?
Gemini introduces Projects and folders for conversation organization, similar to ChatGPT. It also includes Workspace synthesis and memory features.
What are the key updates in Gemini 3.1?
Gemini 3.1 includes Pro and Flash Live agents, video capabilities, and a free Veo tool for videos. It supports 750M monthly active users and 400M weekly active users.
How powerful is Gemma 4?
Gemma 4 offers GPT-5 level performance on devices, with edge agents and local speed via Ollama/MLX quants on Android/iOS. Demis Hassabis hyped it as exceptional for on-phone use, with Hermes benchmarks.
What agent features does Gemini provide?
Gemini has a built-in Agent mode as an alternative to OpenClaw, including TurboQuant and Agent Designer. It supports local runs like Hermes agents on MacBook Air M4.
What integrations does Gemini have with other Google products?
Gemini integrates with Workspace, Search Live, and Siri on iOS 27. It also features post-ChatGPT error tolerance, noted by Gary Marcus.
What licensing and availability options exist for Gemma 4?
Gemma 4 is open-source under Apache 2.0, available in sizes like 2B, 4B, 26B, 31B. It runs locally on setups like Mac mini with Ollama.
How does Gemma 4 perform compared to competitors?
Gemma 4 lags behind some Chinese models but excels in edge and on-device AI with NVIDIA support. It's purpose-built for advanced reasoning and agentic workflows.
What free tools are associated with Gemini?
A free Google AI tool for videos via vids.new is highlighted, alongside Gemma 4's open models. These enable local, cloud-optional AI.
Gemini Projects/folders for convo org like ChatGPT; 3.1 Pro/Flash Live agents/Vids; Gemma 4 edge agents/local speed/TPS Hermes benchmarks/Demis Hassabis repost hype (GPT-5 lvl on-phone); Workspace synthesis/memory; Agent mode native alt; Search Live/Siri; 750M MAU/400M WAU; post-ChatGPT error tolerance.