DeepMind 在 2026年10月 发布的内容

Introducing the DeepMind Institute

DeepMind unveiled the DeepMind Institute (DMI), a new platform for interdisciplinary research and debate on the safe development and governance of artificial general intelligence (AGI). Led by Shane L…

Artificial symbiotic intelligence: Agents, AGI and the orchestration of many minds

DeepMind proposes that AGI may emerge not from a single model but from orchestrated societies of AI agents, tools, and humans, where intelligence is a social phenomenon. The shift from isolated machin…

The case for reasoning transparency

DeepMind highlights the critical role of Chain of Thought (CoT) reasoning in monitoring AI models for misalignment and safety, warning that transparency is not guaranteed in future models. The post em…

Gemini platform

Gemini platform Build, deploy, and govern enterprise-ready AI agents and apps Try Gemini platform

Gemini 3.1 Flash image

DeepMind introduced Gemini 3.1 Flash Image, a high-efficiency image generation model optimized for speed and low latency, positioned as a counterpart to the higher-end Gemini 3 Pro Image. It supports …

Gemini 3.8 Flash-Lite TTS

DeepMind launched Gemini 3.8 Flash-Lite TTS, a cost-efficient text-to-speech model replacing its predecessor for production workloads. The model supports over 100 languages, offers low latency, and ma…

Gemini 3.8 Flash TTS

Google DeepMind introduced Gemini 3.8 Flash TTS, a new flagship text-to-speech model delivering studio-grade voice fidelity, expressive acting, authentic regional accents, and long-form multi-turn sta…

Gemma models overview

DeepMind introduced Gemma, a family of lightweight, state-of-the-art open models built using the same research and technology as the Gemini models. Available in multiple variants for general and speci…

EmbeddingGemma 2: an open, lightweight multimodal embedding model

DeepMind introduced EmbeddingGemma 2, a new open-source, lightweight multimodal embedding model designed for efficiency and versatility. The model supports both text and image inputs, enabling develop…

Gemini Nano Banana 2.1

DeepMind released Gemini Nano Banana 2.1, an update to its high-efficiency image generation model, improving visual quality, prompt adherence, multi-turn consistency, and text rendering while maintain…

EmbeddingGemma 2: an open, lightweight multimodal embedding model October 2026 Models Learn more