GEMINI.md

Evidence (13)

• 8The New Stack·9/24/2026product_launch
  Google adds self-serve custom voice creation to Gemini TTS

  Google has released Gemini 3.8 Flash TTS and Flash-Lite TTS with new self-serve voice creation features in the Gemini API and Google AI Studio. Developers can now design voices from descriptions or replicate a speaker’s voice with consent, save reusable voice IDs, and use content credentials and SynthID to mark generated audio.

• 9Ars Technica AI·9/9/2026security
  U.S. Accuses Six Chinese AI Firms of Copying Frontier Models

  U.S. agencies say DeepSeek, Moonshot AI, Alibaba, MiniMax, StepFun, and Z.AI have been conducting industrial-scale distillation attacks against frontier AI models from U.S. companies since at least late 2024. The agencies allege the firms likely operated with Chinese government awareness, citing use of capabilities extracted from models such as Claude, GPT, Gemini, and Grok.

• 7Google DeepMind·9/1/2026model_release
  Gemini Adds Agentic Video Understanding

  Google DeepMind introduced agentic video understanding in Gemini, aimed at enabling the model to reason over video content and act on it more effectively. The announcement positions Gemini as a stronger multimodal system for video analysis and related agent workflows.

• 6Google DeepMind·8/27/2026model_release
  Gemini Omni 1.1 Flash adds finer control for builders

  Google DeepMind says Gemini Omni 1.1 Flash gives developers more control when building with the model. The update appears focused on improving how the model can be directed and used in applications, rather than introducing a completely new product line.

• 6InfoQ·8/27/2026product_launch
  Google Cloud Launches AI Agents for Database Operations

  Google Cloud has introduced AI-powered Database Operations Agents to simplify database lifecycle management. The initial agents focus on onboarding, observability, troubleshooting, performance optimization, and tuning across services such as AlloyDB, Bigtable, and Spanner.

• 7Google DeepMind·8/27/2026research
  Google DeepMind pilots double-blind AI evaluations

  Google DeepMind says it has piloted the world’s first double-blind evaluation of a proprietary frontier AI model. The approach uses Confidential Space in Google Cloud to keep both the model weights and external benchmark prompts private, with partners including the Singapore AI Safety Institute, OpenMined, AVERI, and MLCommons.

• 2Hacker News·8/26/2026open_source
  YouTube Format IDs reference

  This Hacker News item links to a gist that catalogues YouTube format IDs used in video streams and downloads. It appears to be a reference resource for developers or tools that inspect or select YouTube media formats, rather than a product announcement or major platform change.

• 7The New Stack·8/26/2026product_launch
  Google Launches Gemini Enterprise for Legal

  Google Cloud introduced Gemini Enterprise for Legal, a purpose-built agentic AI offering designed to automate legal workflows such as contract review, regulatory monitoring, document drafting, and data discovery. The launch highlights a broader enterprise AI trend: specialization can come from agents, integrations, governance, and platform layers, not only from a proprietary foundation model.

• 5The New Stack·8/19/2026research
  Coding agents are getting formal onboarding files

  The article argues that companies are increasingly creating dedicated onboarding files for coding agents, such as AGENTS.md, CLAUDE.md, and GEMINI.md, to improve build and task success. It contrasts this rapid optimization for agents with the often neglected documentation and onboarding experience provided to human developers, while noting broader lessons about batching, context switching, and software delivery.

• 7Hacker News·8/16/2026research
  Models Are Deliberately Trading Facts for Reasoning

  The article argues that frontier AI labs are optimizing models to improve reasoning and benchmark performance while reducing the amount of factual knowledge stored in parameters. It cites strong math results but weak factual recall and higher hallucination rates, framing this as a deliberate trade-off toward lightweight, procedure-focused models.

• 8Ars Technica AI·7/30/2026model_release
  Google unveils Gemini Robotics 2.0 for more dexterous robots

  Google DeepMind has introduced Gemini Robotics 2.0, a new set of robotics AI models aimed at improving robot dexterity, situational awareness, and multi-robot collaboration. The update includes an upgraded embodied reasoning model and a publicly available sub-model for developers, with integration into the Gemini Live API.

• 7Ars Technica AI·7/21/2026model_release
  Google launches Gemini 3.6 Flash and a cybersecurity model

  Google introduced three new AI models, including Gemini 3.6 Flash and its first Gemini model aimed at cybersecurity. Gemini 3.5 Flash has been deprecated, while the delayed Gemini 3.5 Pro is still not available and Gemini 4 was only teased. The company says 3.6 Flash improves coding and multimodal performance based on user feedback.

• 3Google DeepMind·5/17/2026model_release
  Google DeepMind Introduces Gemini Omni

  Google DeepMind has announced Gemini Omni. The post provides only the introduction, so no further technical details about capabilities, use cases, or availability are included in the source text.