@GoogleDeepMind

Thomas Wolf is excited about the extension of Terminal-Bench to scientific fields, known as Terminal-Bench Science. This benchmark evaluates AI models' ability to control tools via the command line to achieve scientific goals. It's open for contributions of real scientific workflows until August 2026, aiming to improve AI models' assistance in research work.

Pinecone launches full text search with multi-modal indexing for large-scale text and image data.

Google DeepMind's AlphaEvolve has been accelerating progress in various fields through Gemini technology.

Philipp Schmid利用GoogleDeepMind的Gemini 3.1 Pro观看cursor_ai SDK的发布视频并生成制作脚本，随后使用Remotion无提示重现视频，展示其视频理解能力。

Philipp Schmid展示了一个使用@stripe Link重建的示例，通过1个指令和15分钟完成相同工作流程，展示了AI辅助视频创作的能力。

Patrick Loeber宣布Gemma功能已集成到@geminicli中，用户可通过`gemini gemma setup`和`gemini gemma start`命令体验。此更新伴随Gemini CLI v0.40.0发布，重点改进包括分层内存、本地Gemma路由及简化UI。

GoogleDeepMind showcased a real-time interactive demo of the Genie model, where users define scenes and characters by selecting orbs to generate a navigable virtual world.

Google AI Devs built real AI products within a weekend after Google I/O, but the post lacks technical depth or reusable methodology, focusing only on community energy.

GoogleDeepMind's 8 models dominate the Text Arena Pareto curve, with Gemini 3.5 Flash's pricing strategy reshaping the price-performance competition landscape, as only 4 labs maintain top-tier performance in their respective price brackets.

This is a promotional message about a technical conference, announcing that Qdrant will host technical talks on vector search, AI memory, context engineering, and more.

Recraft's V4.1 Utility Pro has ranked #7 on Design Arena's 2026 image generator leaderboard in graphic design and #9 on Image Arena with Elo 1243 within one week of launch, placing the company among top 5 image generation labs alongside OpenAI, GoogleDeepMind, LumaLabsAI, and bfl_ml.