# Google AI Services – Research (September 2025) This report aggregates public information via MCP Omnisearch (Tavily). Validate details against official Google sources before relying in production. ## AgentSpace - [Google Agentspace](https://cloud.google.com/products/agentspace) - Google Agentspace is the launch point for enterprise-ready AI agents, helping increase employee productivity for complex tasks with one single prompt. - [Announcing the Agent2Agent Protocol (A2A)](https://developers.googleblog.com/en/a2a-a-new-era-of-agent-interoperability/) - The A2A protocol will allow AI agents to communicate with each other, securely exchange information, and coordinate actions on top of various enterprise - [What's new in Google Workspace? - Devoteam](https://www.devoteam.com/event/whats-new-in-google-workspace-q3-2025/) - Gain expert insights on the new features released in Google Workspace from July to September 2025 by registering for this online event. - [Google Agentspace enables the agent-driven enterprise](https://cloud.google.com/blog/products/ai-machine-learning/google-agentspace-enables-the-agent-driven-enterprise) - # Scale enterprise search and agent adoption with Google Agentspace 3. Deploy Google-built agents such as our new Deep Research and Idea Generation agents to help employees generate and validate novel business ideas, synthesize dense information, and more That's what we're bringing to enterprises with Google's AI-powered multimodal search capabilities in Agentspace, helping cus - [What Is Google Agentspace And How Does It Work? - Stewart Gauld](https://stewartgauld.com/what-is-google-agentspace/) - Yes, Google Agentspace lets you create and interact with advanced AI agents to manage tasks, engage in research, automate workflows, and boost efficiency.**But that’s not all!** Created by**_Google Cloud,_****Google Agentspace**is a**multimodal search and AI agent workspace built for you and your team.** With so many impressive AI tools on the market today, you may be wondering ## Co-Scientist - [Accelerating scientific breakthroughs with an AI co-scientist](https://research.google/blog/accelerating-scientific-breakthroughs-with-an-ai-co-scientist/) - [Jump to Content](https://research.google/blog/accelerating-scientific-breakthroughs-with-an-ai-co-scientist/#page-content) * [](https://twitter.com/intent/tweet?text=https%3A//research.google/blog/accelerating-scientific-breakthroughs-with-an-ai-co-scientist/ "Share on Twitter") * [](https://www.facebook.com/sharer/sharer.php?u=https%3A//research.google/blog/accelerating-s - [Accelerating scientific discovery with AI-powered empirical software](https://research.google/blog/accelerating-scientific-discovery-with-ai-powered-empirical-software/) - [Jump to Content](https://research.google/blog/accelerating-scientific-discovery-with-ai-powered-empirical-software/#page-content) * [](https://twitter.com/intent/tweet?text=https%3A//research.google/blog/accelerating-scientific-discovery-with-ai-powered-empirical-software/ "Share on Twitter") * [](https://www.facebook.com/sharer/sharer.php?u=https%3A//research.google/blog/ - ["I think it will change science," AI just did in 2 days what researchers ...](https://www.futura-sciences.com/en/i-think-it-will-change-science-ai-just-did-in-2-days-what-researchers-took-10-years-to-find_19911/) - September 10, 2025 by Edward Back and Futura Team. For a decade ... Then came Google's new AI system, Co-scientist. In just 48 hours, it - [Google's top AI scientist says 'learning how to learn' will be next ...](https://techxplore.com/news/2025-09-google-ai-scientist-generation-skill.html) - ![Google's top AI scientist says ‘learning how to learn’ will be next generation's most needed skill](https://scx1.b-cdn.net/csz/news/800a/2025/googles-top-ai-scienti.jpg "Demis Hassabis, CEO of Google's artificial intelligence research company DeepMind, right, and Greece's Prime Minister Kyriakos Mitsotakis discuss the future of AI, ethics and democracy during an event at the - [Google Unveils AI Scientist That Could Transform Research - HPCwire](https://www.hpcwire.com/2025/02/26/google-unveils-ai-scientist-that-could-transform-research/) - The Google AI Co-Scientist is built on its Gemini 2.0 AI model, and is specifically designed to “aid scientists in creating novel hypotheses and research plans.” The tool works much like any other AI chatbot where the user enters a research goal using natural language, and the AI would reply. The upcoming Doudna supercomputer at the National Energy Research Scientific Computing ## Project Mariner - [Google rolls out Project Mariner, its web-browsing AI agent](https://techcrunch.com/2025/05/20/google-rolls-out-project-mariner-its-web-browsing-ai-agent/) - # Google rolls out Project Mariner, its web-browsing AI agent Google announced during Google I/O 2025 that it’s rolling out Project Mariner, the company’s experimental AI agent that browses and uses websites, to more users and developers. Google also says it’s bringing Project Mariner’s capabilities to the Gemini API and Vertex AI, allowing developers to build out applications - [Project Mariner - Google DeepMind](https://deepmind.google/models/project-mariner/) - Use natural language to assign AI agents to handle time-consuming tasks, like research, planning, and data entry. Project Mariner observes what’s displayed in the browser. Once agents have learned how to do a task, they can try to replicate the same workflow in the future with minimal input — freeing up even more of your time. Project Mariner uses information from a resume to f - [AI in Search: Going beyond information to intelligence - Google Blog](https://blog.google/products/search/google-search-ai-mode-update/) - [Skip to main content](https://blog.google/products/search/google-search-ai-mode-update/#jump-content) [Share](https://blog.google/products/search/google-search-ai-mode-update/) * [Search](https://blog.google/products/search/) * [Search](https://blog.google/products/search/) * [AI](https://blog.google/technology/ai/) * [AI](https://blog.google/technology/ai/) * [Searc - [How Will Google's Project Mariner Redefine User Testing? - Userlytics](https://www.userlytics.com/resources/blog/how-will-googles-project-mariner-redefine-usability-and-user-testing/) - Before all Chrome users are given access to the helpful AI agent, let’s explore what Google’s Project Mariner is and how it will redefine usability and user testing in the near future. Google’s Project Mariner is the experimental testing of a Gemini 2.0-powered agent (Mariner) that can understand the contents of a user’s Chrome browser, allowing it to navigate websites much lik - [Google's Project Mariner: The AI Agent That Browses the Web](https://designforonline.com/googles-project-mariner-the-ai-agent-that-browses-the-web/) - Discover Google's Project Mariner. An AI Agent that browses the web like a human. See real world AI Automation for business use cases ## Lyria - [Google Lyria 2: AI Music Generation Made Simple \| Scenario](https://help.scenario.com/en/articles/google-lyria-the-essentials/) - Google Lyria 2 represents Google's latest advancement in AI-powered music generation ... September 10, 2025. On this page. 1. Overview of Google - [Lyria \| AI Music Generator \| Generative AI on Vertex AI - Google Cloud](https://cloud.google.com/vertex-ai/generative-ai/docs/music/generate-music) - * Lyria Vertex AI model versions and lifecycle # Lyria | AI Music Generator bookmark\_borderbookmark Stay organized with collections Save and categorize content based on your preferences. You can use Lyria to generate new instrumental music tracks from a text prompt Lyria offers generative AI features for music creation. Lyria generates instrumental music from text prompts. 1. - [Music generation using Lyria RealTime \| Gemini API](https://ai.google.dev/gemini-api/docs/music-generation) - # Music generation using Lyria RealTime generation model. ## How music generation works Lyria RealTime music generation uses a persistent, bidirectional, ## Generate and control music configuration using `session.set_music_generation_config`, starts the music client.aio.live.music.connect(model='models/lyria-realtime-exp') as session, await session.set_weighted_prompts( await s - [Lyria - Google DeepMind](https://deepmind.google/models/lyria/) - Lyria 2 delivers **high-fidelity music** and **professional-grade audio**, capturing subtle nuances across a range of genres and intricate compositions. * Music AI Sandbox Lyria 2 is an advanced AI model, developed with input and insights from musicians and producers. Lyria 2 can broaden musical understanding and help people explore new creative avenues. ## Music AI Sandbox Pow - [Google brings a music-generating AI model to its API with Lyria ...](https://techcrunch.com/2025/05/20/google-brings-a-music-generating-ai-model-to-its-api-with-lyria-realtime/) - AI # Google brings a music-generating AI model to its API with Lyria RealTime Google announced at Google I/O 2025 that it is making the AI model that powers its experimental music production app, MusicFX DJ, available via an API. The model, Lyria RealTime, is now in Google’s Gemini API and AI Studio platform. The launch of Lyria RealTime via API comes amidst an explosion of AI- ## Chirp - [Text-to-Speech release notes - Google Cloud](https://cloud.google.com/text-to-speech/docs/release-notes) - Chirp 3: HD voices now supports 8 new speakers in 31 new locales: ar-XA, bn-IN, cmn-CN, de-DE, en-AU, en-GB, en-IN, en-US, es-ES, es-US, fr-CA, fr-FR, gu-IN, hi-IN, id-ID, it-IT, ja-JP, kn-IN, ko-KR, ml-IN, mr-IN, nl-NL, pl-PL, pt-BR, ru-RU, sw-KE, ta-IN, te-IN, th-TH, tr-TR, and vi-VN. Cloud Text-to-Speech now offers updated Journey voices with an additional speaker, en-us-Jou - [Chirp 3: HD voices \| Cloud Text-to-Speech API](https://cloud.google.com/text-to-speech/docs/chirp3-hd) - Text-to-Speech Chirp 3: HD voices represent the latest generation of Text-to-Speech technology. Discover how to use Chirp 3: HD voices to synthesize speech. """Synthesizes speech from a stream of input text.""" # See https://cloud.google.com/text-to-speech/docs/voices for all voices. Creating engaging and natural-sounding audio from text requires understanding the nuances of sp - [Google Launches Chirp 3 on Vertex AI to Revolutionize Voices](https://www.openxcell.com/ai-news/google-launches-chirp-3-on-vertex-ai/) - Google Launches Chirp 3 on Vertex AI to Revolutionize Voices - Openxcell * Ready-to-deploy AI Solutions Image 9: Artificial Intelligence menu img-img But now, the tech giant Google has taken a bold step by integrating model **Chirp 3** into its Vertex AI platform. This move solidifies Google’s position in the **Gen AI world**, which further allows developers to craft cuttin - [Google Cloud Launches Chirp 3 Voice Model Integration - AI Business](https://aibusiness.com/language-models/google-cloud-launches-chirp-3-voice-model-integration) - has made its high-definition voice model, Chirp 3, available for developers through the Vertex AI unified machine learning platform. Google announced the release using a Chirp 3 AI voice at a launch event at Google DeepMind’s London headquarters. "Our general vision for AI has been to offer a broad range of models,” said Google Cloud CEO Thomas Kurian. The company is offering c - [Google Cloud's Chirp 3 Ushers in a New Era of Realistic AI ...](https://www.linkedin.com/pulse/google-clouds-chirp-3-ushers-new-era-realistic-audio-malaviarachchi-si60c) - Google Cloud's Chirp 3 Ushers in a New Era of Realistic AI-Generated Audio on Vertex AI Image 2: Google Cloud's Chirp 3 Ushers in a New Era of Realistic AI-Generated Audio on Vertex AI Google Cloud's Chirp 3 Ushers in a New Era of Realistic AI-Generated Audio on Vertex AI Google Cloud has officially launched Chirp 3, its most advanced text-to-audio model to date, making it avai ## Veo3 - [Veo 3 \| Google AI Studio](https://aistudio.google.com/models/veo-3) - Our state-of-the-art video generation model Video generation meets audio Bring your generated videos to life with native audio generation Veo 3 pairs audio with the context of what you’re generating so you can bring your stories to life in a way you couldn’t before Now you can generate video with audio from text or an image, making it perfect for building apps that need to prod - [Google Photos upgrades its image-to-video feature with Veo 3](https://techcrunch.com/2025/09/04/google-photos-adds-ai-video-generation-with-veo-3/) - [Skip to content](https://techcrunch.com/2025/09/04/google-photos-adds-ai-video-generation-with-veo-3/#wp--skip-link--target) * [Latest](https://techcrunch.com/latest/) * [Startups](https://techcrunch.com/category/startups/) * [Venture](https://techcrunch.com/category/venture/) * [Apple](https://techcrunch.com/tag/apple/) * [Security](https://techcrunch.com/category/s - [Veo 3 comes to Google Photos. Try it in the new Create tab.](https://blog.google/products/photos/google-photos-create-tab-editing-tools/) - With Photo to video, you can already turn your images into fun, short clips, and starting today we’re enhancing that feature with our state-of-the-art video generation model, Veo 3. You can find this powerful new tool, along with other generative AI features, in our new Create tab**.** Now available in the U.S., the Create tab is your central hub for creativity in Google Photos - [Veo 3 AI Video Generator with Audio \| veo3.ai](https://veo3.ai/) - [@nmatares](https://x.com/nmatares/status/1924931844879134804) [](https://x.com/nmatares/status/1924931844879134804) [@hhm](https://x.com/hhm/status/1924982194457690176) [](https://x.com/hhm/status/1924982194457690176) The person comments out loud on what he sees, and what surprises him the most" [#veo3](https://x.com/hashtag/veo3) [Read 145 replies](https://x.com/hhm/status/19 - [Gemini AI video generator powered by Veo 3](https://gemini.google/overview/video-generation/) - Create high-quality, 8-second videos with Veo 3, our latest AI video generator. Create videos with sound using our video generation model that maintains high-quality while optimizing for speed. Create 8 second videos Veo 3 Create high-quality, 8-second videos with sound using our state-of-the-art video generation model. Create 8 second videos Yes, you can create and share video ## Imagen4 - [Announcing Imagen 4 Fast and the general availability of ...](https://developers.googleblog.com/en/announcing-imagen-4-fast-and-imagen-4-family-generally-available-in-the-gemini-api/) - # Announcing Imagen 4 Fast and the general availability of the Imagen 4 family in the Gemini API We're excited to announce that Imagen 4, our most advanced text-to-image model, is now generally available in the Gemini API and Google AI Studio. * **[New] - Imagen 4 Fast**: Ideal for rapid image generation and high-volume tasks, this model offers incredible speed at an accessible - [Imagen 4 Ultra Generate \| Generative AI on Vertex AI](https://cloud.google.com/vertex-ai/generative-ai/docs/models/imagen/4-0-ultra-generate-001) - * Generative AI on Vertex AI # Imagen 4 Ultra Generate bookmark\_borderbookmark Stay organized with collections Save and categorize content based on your preferences. Imagen 4 is our newest line of image generation models. This page documents the capabilities and features of `imagen-4.0-ultra-generate-001`. | Model ID | `imagen-4.0-ultra-generate-001` | | | Capabilities | * Sup - [Imagen 4 is Google's newest AI image generator](https://techcrunch.com/2025/05/20/imagen-4-is-googles-newest-ai-image-generator/) - AI # Imagen 4 is Google’s newest AI image generator Google is rolling out a new image-generating AI model, Imagen 4, that the company claims delivers higher-quality results than its previous image generator, Imagen 3. Unveiled at Google I/O 2025 on Tuesday, Imagen 4 is capable of rendering “fine details” like fabrics, water droplets, and animal fur, Google says. According to Go - [Imagen 4 is now available in the Gemini API and Google AI ...](https://developers.googleblog.com/en/imagen-4-now-available-in-the-gemini-api-and-google-ai-studio/) - # Imagen 4 is now available in the Gemini API and Google AI Studio We're thrilled to bring Imagen 4, **our best text-to-image model yet**, to paid preview in the Gemini API and for limited free testing in Google AI Studio. **Imagen 4 offers significantly improved text rendering over our prior image models** and pushes the boundaries of text-to-image generation quality. This is - [Imagen](https://deepmind.google/models/imagen/) - Our leading text-to-image model, engineered for creativity Imagen 4 is optimized for creativity, generating images with up to 2k resolution. Express your ideas like never before — with Imagen, creativity has no limits. In tests, people prefer the latest version of Imagen to previous models — and other leading text-to-image models. Imagen 4 is leading text-to-image — but we’re s ## Multi-modal Streaming - [Google models \| Generative AI on Vertex AI](https://cloud.google.com/vertex-ai/generative-ai/docs/models) - MedLM models. Caution: MedLM is deprecated. Access to MedLM will no longer be available on or after September 29, 2025. - [Supported models \| Generative AI on Vertex AI - Google Cloud](https://cloud.google.com/vertex-ai/generative-ai/docs/provisioned-throughput/supported-models) - Provisioned Throughput only supports models that you call directly Provisioned Throughput to make API calls to a model, you must use the for Google models that support Provisioned Throughput. | Gemini 2.5 Flash-Lite Latest supported version: `gemini-2.5-flash-lite` | 8070 | Tokens | 1 | 1 input text token = 1 token 1 input image token = 1 token 1 input video token = 1 token - [Learn about supported models \| Firebase AI Logic - Google](https://firebase.google.com/docs/ai-logic/models) - Our multimodal model that supports low-latency realtime streaming of multimodal inputs and outputs. 1 Additional model names support the Live API depending on - [Gemini flash Live API docs chaos sorted out: - Documentation](https://discuss.ai.google.dev/t/gemini-flash-live-api-docs-chaos-sorted-out/80120) - Successfully implementing the desired real-time audio interaction requires pinpointing the exact Gemini 2.0 Flash model variant on Vertex AI that supports the Multimodal Live API and its Text-to-Speech (TTS) output capability. The evidence confirms that the **`gemini-2.0-flash` model on Vertex AI supports the Multimodal Live API (via the `BidiGenerateContent` service)** necessa - [Hi Gemini, Secure the Room - Google Cloud - Medium](https://medium.com/google-cloud/hi-gemini-secure-the-room-01903d5541f6) - This project demonstrates how modern AI APIs can be integrated into real-time systems with careful attention to latency and performance. By --- Notes: - Include API endpoints, auth, and pricing after validating vendor docs. - Mark services as production/preview based on official release notes. --- ## ✅ Official Verification (Auto-collected) Verified: 2025-09-13T23:26:58.523Z ### AgentSpace - Authentication: Likely Google Cloud authentication (verify) - Notes: No explicit quotas found in summaries (see links) ### Project Mariner - Authentication: Likely Google Cloud authentication (verify) - Notes: No explicit quotas found in summaries (see links) ### Veo3 - Authentication: Likely Google Cloud authentication (verify) - Notes: No explicit quotas found in summaries (see links) ### Imagen4 - Authentication: Likely Google Cloud authentication (verify) - Notes: No explicit quotas found in summaries (see links) ### Lyria - Authentication: Likely Google Cloud authentication (verify) - Notes: No explicit quotas found in summaries (see links) ### Chirp - Authentication: Likely Google Cloud authentication (verify) - Notes: No explicit quotas found in summaries (see links) ### Co-Scientist - Authentication: Likely Google Cloud authentication (verify) - Notes: No explicit quotas found in summaries (see links) ### Streaming API - Authentication: Likely Google Cloud authentication (verify) - Notes: No explicit quotas found in summaries (see links) --- ## 🧭 Capability Matrix (Repo vs Service) | Service | Code Files | Status | |---|---|---| | AgentSpace | src/services/google-services/agent-space-manager.ts | Conceptual (adapters/examples) | | Project Mariner | src/services/google-services/mariner-automation.ts | Conceptual (adapters/examples) | | Veo3 | src/services/google-services/veo3-video-generator.ts | Conceptual (adapters/examples) | | Imagen4 | src/services/google-services/imagen4-generator.ts | Conceptual (adapters/examples) | | Lyria | src/services/google-services/lyria-music-composer.ts | Conceptual (adapters/examples) | | Chirp | src/services/google-services/chirp-audio-processor.ts | Conceptual (adapters/examples) | | Co-Scientist | src/services/google-services/co-scientist-research.ts | Conceptual (adapters/examples) | | Streaming API | src/services/google-services/enhanced-streaming-api.ts
src/streaming/enhanced-streaming-api.ts | Conceptual (adapters/examples) | --- ## 🧩 Implementation Guides (Verified) _Puppeteer-verified extracts from official docs. Always confirm with latest docs before deployment._ ### Imagen4 **Official References:** - (none) **Authentication:** See docs **Rate Limits/Quotas:** See docs **Recommended SDK Imports:** @google-cloud/aiplatform ### Veo3 **Official References:** - (none) **Authentication:** See docs **Rate Limits/Quotas:** See docs **Recommended SDK Imports:** @google-cloud/aiplatform ### Streaming API **Official References:** - (none) **Authentication:** See docs **Rate Limits/Quotas:** See docs **Recommended SDK Imports:** @google-cloud/aiplatform, @google/generative-ai --- ## 🧩 Implementation Guides (Verified) _Puppeteer-verified extracts from official docs. Always confirm with latest docs before deployment._ ### Imagen4 **Official References:** - https://cloud.google.com/vertex-ai/generative-ai/docs/image/overview: https://cloud.google.com/vertex-ai/generative-ai/docs/image/overview - https://cloud.google.com/vertex-ai/generative-ai/docs/image/text-image: https://cloud.google.com/vertex-ai/generative-ai/docs/image/text-image **Authentication:** See docs — likely Google Cloud auth **Rate Limits/Quotas:** No explicit quotas detected on page **Recommended SDK Imports:** @google-cloud/aiplatform ### Veo3 **Official References:** - https://cloud.google.com/vertex-ai/generative-ai/docs/video/overview: https://cloud.google.com/vertex-ai/generative-ai/docs/video/overview **Authentication:** See docs — likely Google Cloud auth **Rate Limits/Quotas:** No explicit quotas detected on page **Recommended SDK Imports:** @google-cloud/aiplatform ### Streaming API **Official References:** - https://cloud.google.com/vertex-ai/generative-ai/docs/multimodal/realtime: https://cloud.google.com/vertex-ai/generative-ai/docs/multimodal/realtime **Authentication:** See docs — likely Google Cloud auth **Rate Limits/Quotas:** No explicit quotas detected on page **Recommended SDK Imports:** @google-cloud/aiplatform, @google/generative-ai --- ## 🧩 Implementation Guides (Verified) _Puppeteer-verified extracts from official docs. Always confirm with latest docs before deployment._ ### Imagen4 **Official References:** - https://cloud.google.com/vertex-ai/generative-ai/docs/image/overview: https://cloud.google.com/vertex-ai/generative-ai/docs/image/overview - https://cloud.google.com/vertex-ai/generative-ai/docs/image/text-image: https://cloud.google.com/vertex-ai/generative-ai/docs/image/text-image **Authentication:** See docs — likely Google Cloud auth **Rate Limits/Quotas:** No explicit quotas detected on page **Recommended SDK Imports:** @google-cloud/aiplatform ### Veo3 **Official References:** - https://cloud.google.com/vertex-ai/generative-ai/docs/video/overview: https://cloud.google.com/vertex-ai/generative-ai/docs/video/overview **Authentication:** See docs — likely Google Cloud auth **Rate Limits/Quotas:** No explicit quotas detected on page **Recommended SDK Imports:** @google-cloud/aiplatform ### Streaming API **Official References:** - https://cloud.google.com/vertex-ai/generative-ai/docs/multimodal/realtime: https://cloud.google.com/vertex-ai/generative-ai/docs/multimodal/realtime **Authentication:** See docs — likely Google Cloud auth **Rate Limits/Quotas:** No explicit quotas detected on page **Recommended SDK Imports:** @google-cloud/aiplatform, @google/generative-ai