News
Product launches, model releases, and platform announcements — what changed and what it means.
Gemini 3.5 Flash gets built-in computer use
Google made computer use a built-in tool in Gemini 3.5 Flash. Developers can now build agents that click, type and inspect browser, mobile and desktop interfaces through Gemini API or Agent Platform.
Gemini 3.5 Pro reportedly slips to July
Gemini 3.5 Pro has reportedly slipped from June to July while Google gathers tester feedback. The delay matters because agentic coding has become the frontier-model battleground.
DiffusionGemma tests a faster path for open-weight text models
Google’s experimental DiffusionGemma trades some polish for speed, testing whether open-weight text models can feel more interactive on local and dedicated GPUs.
Google proposes Open Knowledge Format for agent context
Google Cloud introduced OKF, a markdown-and-YAML format for portable agent context. It targets data teams that need AI agents to share metadata without a proprietary catalog lock-in.
DiffusionGemma tests a faster path for open-weight text models
Google’s experimental DiffusionGemma trades some polish for speed, testing whether open-weight text models can feel more interactive on local and dedicated GPUs.
Gemini 3.5 Live Translate turns real-time speech translation into a developer surface
Google’s Gemini 3.5 Live Translate is rolling into the Live API, AI Studio, Translate, and Meet, with 70+ language detection and SynthID-watermarked audio.
Google ships Gemma 4 12B for local multimodal agents
Gemma 4 12B is a dense encoder-free multimodal open model for laptops. It supports text, image, audio and video inputs with a 256K context window.
Gemma 4 12B gets local Mac apps and LiteRT-LM serving
Gemma 4 12B now runs through Google AI Edge Gallery, Eloquent, and LiteRT-LM serving. The launch targets local multimodal agents on laptops with 16GB memory.
Gemini 3.5 Flash arrives with agentic intelligence
Google launches the 3.5 family. Flash ships today across the Gemini app, Search, Antigravity, and the API; Pro lands in June.