## **August 2025** ## Google - **[Gemini 2.5 Flash Image (aka Nano-Banana) Released:](https://developers.googleblog.com/en/introducing-gemini-2-5-flash-image/)** Google has unveiled its new state-of-the-art image model, Gemini 2.5 Flash Image. - **[NotebookLM Video Overview Supports 80 Languages:](https://blog.google/technology/google-labs/notebook-lm-audio-video-overviews-more-languages-longer-content/)** NotebookLM has been updated to support Video and Audio Overviews in over 80 languages. This feature allows users worldwide to generate structured, narrated slideshows and detailed audio summaries from their notes, PDFs, and images. - **[Gemini CLI GitHub Actions and Custom Commands:](https://github.com/google-gemini/gemini-cli)** New GitHub Actions for Gemini CLI are now in beta. They allow the AI to autonomously assist with development workflows like triaging issues, reviewing pull requests, and engaging in conversations. - **[Gemini Code Assist 2.0 Agent Mode Widely Available:](https://developers.google.com/gemini-code-assist/resources/release-notes)** The agent mode for Gemini Code Assist is now broadly available, offering full remote codebase awareness, the ability to get suggestions from documentation, and new tools for multi-file editing and code diffs within the chat interface. - **[Jules Exits Beta:](https://blog.google/technology/google-labs/jules-now-available/)** Google's asynchronous AI coding agent, Jules, has officially exited beta. The tool, powered by Gemini 2.5 Pro, integrates with GitHub to autonomously fix and update code. - **[Gemini Adds Memory & Temporary Chats:](https://blog.google/products/gemini/temporary-chats-privacy-controls/)** Google has rolled out **Personal Context**, a new memory feature for Gemini that automatically remembers details from past chats. It also introduced **Temporary Chats**, an incognito-like mode for one-off conversations that are not saved. - **[Google adds Storybook feature in Gemini](https://gemini.google/overview/storybook/)** - You can now easily create personalized, illustrated stories about anything with read-aloud narration. - **[Gemma 3 270M Released:](https://developers.googleblog.com/en/introducing-gemma-3-270m/)** Google released **Gemma 3 270M**, a new compact and power-efficient open model. It is designed for on-device and research applications and features strong instruction-following capabilities for its size. - **[Google AI Pro for Students:](https://gemini.google/students/)** Google is offering college students one year of its **AI Pro** plan for free. The plan includes access to Gemini 2.5 Pro, Veo 3, and a new **Guided Learning** mode for the Gemini chatbot designed to promote critical thinking. - **[Veo 3 Integrated into Google Vids:](https://blog.google/feed/new-ai-vids-no-cost-option/)** Google's video generation model, **Veo 3**, is now integrated into **Google Vids**, allowing users to generate video content directly within the application. ## Microsoft - **[Microsoft Unveils New Homegrown AI Models:](https://microsoft.ai/news/two-new-in-house-models/)** Microsoft's AI division, led by Mustafa Suleyman, unveiled two new AI models: **MAI-Voice-1** for efficient speech and **MAI-1-preview**, a text-based model to power future versions of Copilot. - **[Microsoft Launches Copilot Agent Academy:](https://github.com/microsoft/agent-academy)** Microsoft launched the **Copilot Agent Academy**, a new training program for developers and partners focused on building and deploying AI agents using Copilot Studio and other Microsoft AI tools. - **[Copilot in Excel Gets New Functions:](https://techcommunity.microsoft.com/blog/microsoft365insiderblog/bring-ai-to-your-formulas-with-the-copilot-function-in-excel/4443487)** Microsoft Excel's Copilot integration now includes a "Function-Finder" that can automatically write complex formulas based on natural language descriptions and new data visualization tools. - **[Microsoft Incorporates GPT-5:](https://news.microsoft.com/source/features/ai/openai-gpt-5/)** Microsoft announced the deep integration of **GPT-5** into Copilot for Microsoft 365, enabling more advanced reasoning and multi-modal understanding for enterprise users. - **[Copilot Launches on Samsung TVs/Monitors:](https://news.samsung.com/global/samsung-brings-microsoft-copilot-to-2025-tvs-and-monitors-unlocking-smarter-on-screen-experiences)** Samsung has partnered with Microsoft to integrate the **Copilot** AI assistant directly into its 2025 lineup of AI-powered TVs and Smart Monitors. - **[Microsoft Research on Steerable Virtual Scientist:](https://www.microsoft.com/en-us/research/blog/self-adaptive-reasoning-for-science/)** Microsoft Research published a paper on the **Steerable Virtual Scientist**, a self-adaptive reasoning framework for AI agents in scientific discovery. The framework allows the agent to modify its own behavior based on the results of a scientific experiment. ![[Copilot function in Excel.png]] ## OpenAI - **[OpenAI Releases GPT-5:](https://openai.com/index/introducing-gpt-5)** OpenAI officially launched **GPT-5**, its new flagship model. It features a major leap in reasoning and multi-modal understanding and is now the default model for ChatGPT. - **[OpenAI Launches GPT-OSS Open-Source Models:](https://openai.com/open-models/)** OpenAI has introduced **GPT-OSS**, a new family of open-source models available on Hugging Face. The models are smaller, more efficient versions of their larger counterparts, designed for the open-source community. - **[New Voice Model & API for Voice Agents:](https://openai.com/index/introducing-gpt-realtime/)** OpenAI announced an upgraded **Voice Model** and new API functionalities specifically for developers building voice-enabled AI agents, allowing for more natural and low-latency conversations. - **[OpenAI Launches agents.md:](https://openai.com/index/introducing-codex/)** OpenAI introduced **agents.md**, a new online portal for the community to share, discover, and collaborate on AI agent projects, complete with templates and tutorials. - **[OpenAI Opens New Office in India:](https://openai.com/global-affairs/learning-accelerator/)** OpenAI announced the opening of a new office in Bengaluru, India, to tap into the country's AI talent pool and support local startups. - **[Deep Integration of GPT-5 into Salesforce:](https://openai.com/index/salesforce/)** OpenAI announced a deep integration of **GPT-5** into the Salesforce Platform, enhancing CRM and customer service tools with advanced reasoning and conversational AI. ## xAI - [**Grok Code Fast 1**](https://x.ai/news/grok-code-fast-1): We're thrilled to introduce grok-code-fast-1, a speedy and economical reasoning model that excels at agentic coding. - [**Grok Imagine, xAI's new AI image and video generator...**](https://techcrunch.com/2025/08/04/grok-imagine-xais-new-ai-image-and-video-generator-lets-you-make-nsfw-content/): Grok Imagine, xAI's new AI image and video generator, lets you make NSFW content, now available on mobile. ## Anthropic - [**Claude Opus 4.1**](https://www.anthropic.com/news/claude-opus-4-1): Anthropic released **Claude Opus 4.1**, an incremental update to its most powerful model with enhancements in reasoning and instruction-following. - [**Claude Opus 4 and 4.1 can now end a rare subset of conversations**](https://www.anthropic.com/research/end-subset-conversations): Claude Opus 4 and 4.1 have the ability to end a rare subset of conversations, and reference past chats. - [**Piloting Claude for Chrome**](https://www.anthropic.com/news/claude-for-chrome): A Claude extension for Chrome where trusted users can instruct Claude to take actions on their behalf within the browser. - [**Detecting and countering misuse of AI: August 2025**](https://www.anthropic.com/news/detecting-countering-misuse-aug-2025): Our Threat Intelligence report discusses several recent examples of Claude being misused. - [**Anthropic's Claude Gains Memory Feature to Enhance User Chat...**](https://autoblogging.ai/news/ai/anthropics-claude-gains-memory-feature-to-enhance-user-chat-experience/): This new functionality aims to greatly enhance user interactions, allowing Claude to retain and recall information from previous conversations. - [**Anthropic is giving Claude to U.S. government for $1**](https://www.cnbc.com/2025/08/12/anthropic-claude-government-ai.html): As AI race heats up. ## Others - [**Medium is the new large**](https://mistral.ai/news/mistral-medium-3): Mistral Medium 3.1, the frontier-class multimodal model from Mistral was released, improving tone and performance. - [**Comet Plus**](https://www.perplexity.ai/hub/blog/introducing-comet-plus): Comet Plus is a new subscription that gives Perplexity users access to premium content from a group of trusted publishers and journalists. - [**Manus AI Launches Wide Research**](https://manus.im/blog/introducing-wide-research): Deploying 100 agents for comprehensive searches. - [**ElevenLabs Agents now support Chat Mode**](https://elevenlabs.io/blog/elevenlabs-agents-now-support-chat-mode): ElevenLabs introduced Chat Mode, a new capability that lets users build text-only conversational agents. - [**Eleven Music**](https://elevenlabs.io/blog/eleven-music-is-here): Studio-grade music generated with natural language prompts in any style and for countless uses. - [**Grammarly Launches Specialized AI Agents**](https://www.grammarly.com/blog/company/grammarly-launches-ai-agents/): Real-time AI agents provide assistance at every stage of the writing process while maintaining user control and authenticity. - [**Docker Desktop 4.44 Release**](https://www.docker.com/blog/docker-desktop-4-44/): This expanded client support allows Goose and Gemini users to access containerized MCP servers such as GitHub, Postgres, Neo4j, and many others. - [**Cohere raises $500M at $6.8B valuation**](https://cohere.com/blog/august-2025-funding-round): Cohere raises $500M at $6.8B valuation to accelerate enterprise efficiency with agentic AI, and introduced Command A Vision.