Image created with OpenAI gpt-image-1. Image prompt: Single-panel cartoon with loose, hand‑inked lines, bean‑bodied figures, muted flat colors, minimal props, and deadpan humor: Enormous octopus librarian in a corridor of card catalogs, each tentacle handing scrolls marked Search, Maps, Ads, Email to tiny assistant crabs. Large bold title text centered at top: “GOOGLE” Muted colors, flat shading, black ink outlines. 16:9. Caption: “Google rehires its original web crawler.”
“Perplexity has been asked to testify in the Google DOJ case. Our core points: 1. Google should not be broken up. Chrome should remain within and continue to be run by Google. Google deserves a lot of credit for open-sourcing Chromium, which powers Microsoft’s Edge and will also” / X https://x.com/AravSrinivas/status/1914373458982805888
AI Overviews Reduce Clicks by 34.5%
https://ahrefs.com/blog/ai-overviews-reduce-clicks/
“*Taps mic* Is this thing on? 🎙️ We’re excited to introduce new features in Music AI Sandbox, a set of experimental tools for professional musicians. Powered by our latest model Lyria 2, they’re helping singer-songwriters like Isabella Kensington create their next masterpiece. https://x.com/GoogleDeepMind/status/1915421048171856047
“1️⃣ The create feature in Music AI Sandbox enables generation of multiple unique tracks to spark creativity. 🪄 Choose vocal style 🎤, input your lyrics ✍️, fine-tune tempo and key ♯ using the improved advanced controls. https://x.com/GoogleDeepMind/status/1915421050768081285
“Demis Hassabis says today’s AI lacks consciousness, but self-awareness could emerge “implicitly.” Models may soon need to understand “you, self, and other” — the early elements of awareness. They lack imagination now, but in 5–10 years, Hassabis predicts they’ll solve and pose https://x.com/vitrupo/status/1914167730212855978
“Today we’re launching Research, alongside a new Google Workspace integration. Claude now brings together information from your work and the web. https://x.com/AnthropicAI/status/1912192384588271771
“This will be the future! 🤯 At Google Cloud we showed how we think the next Generation of Agents will work, personalized real-time and Multimodal! All powered @GoogleDeepMind Gemini 2.0 Flash and the Live API. Technical Details: 🗣️ Build with Gemini 2.0 Flash LIVE API and ADK. https://x.com/_philschmid/status/1915360039570739283
“Gemini 2.5 Flash just dropped. ⚡ As a hybrid reasoning model, you can control how much it ‘thinks’ depending on your 💰 – making it ideal for tasks like building chat apps, extracting data and more. Try an early version in @Google AI Studio → https://x.com/GoogleDeepMind/status/1912966489415557343
“We were busy over eastern, here are some updates you might have missed, we shipped: – @GoogleDeepMind Gemini 2.5 Flash, dynamic thinking and the best cost-performance Gemini Model released – Context Caching for Gemini 2.0 Flash, 2.5 Pro (2.5 Flash coming), ~75% cost reduction https://x.com/_philschmid/status/1914571503397396956
“📣 It’s here: ask Gemini about anything you see. Share your screen or camera in Gemini Live to brainstorm, troubleshoot, and more. Rolling out to Pixel 9 and Samsung Galaxy S25 devices today and available for all Advanced users on @Android in the Gemini app: https://x.com/GeminiApp/status/1909215393186472380
“Google continues to cook—this time dropping Gemini 2.5 Flash, a hybrid reasoner AI —Matches 04-mini —Strong across reasoning, STEM, and visual reasoning —Includes ‘thinking budget’ to optimize for cost —Available via Google AI Studio, Vertex, Gemini app https://x.com/rowancheung/status/1914201233822253336
“Just announced new versions of Gemma 3 – the most capable model to run just one H100 GPU – can now run on just one *desktop* GPU! Our Quantization-Aware Training (QAT) method drastically brings down memory use while maintaining high quality. Excited to make Gemma 3 even more https://x.com/sundarpichai/status/1913260423622656432
“Google also released a new version of Gemma 3, optimized with ‘Quantization-Aware Training’ This reduces the 27B model’s memory requirements, enabling it to run on consumer GPUs, like the NVIDIA RTX 3090, with maintained performance https://x.com/rowancheung/status/1914201267053777278
“Demis Hassabis appeared on 60 Minutes, sharing new insights into AI: —AGI in 5-10 years, with conscious AI likely in the future —AI fast-tracking drug discovery, potentially eradicating all disease within the next decade —AI ensuring radical abundance https://x.com/rowancheung/status/1914567398885147089
Achieve real-time interaction: Build with the Live API – Google Developers Blog https://developers.googleblog.com/en/achieve-real-time-interaction-build-with-the-live-api/
“We just shipped Tier 3 limits in the Gemini API (2x-6x higher limits), available now for self serve upgrade if you have spent >=$1K with Google Cloud. You can upgrade in AI Studio (on the API key page) to Tier 3 and start scaling with Gemini right now! https://x.com/OfficialLoganK/status/1915119791506915812
Google for Startups AI Academy: American Infrastructure cohort applications open https://blog.google/feed/google-for-startups-ai-academy-america-infrastructure-apply/
“and with @GoogleDeepMind’s 2.5 Flash official blogpost adopting this chart, i think its fair to say that 1) we’ve successfully gotten the industry onboard with @polynoamial’s desire for performance-per-cost thinking, 2) time is probably right to move past LMArena elo going https://x.com/swyx/status/1912959140743586206
“Catch them all! Gemini 2.5 Pro continues to make progress in completing Pokémon! So far It earned 5 badges (next best model only has 3 so far, though with a different agent setup) and has a level 58 Blastoise. 🤯 https://x.com/_philschmid/status/1914199732013985840
“Here’s the first on‑camera look at Google’s Android XR glasses AI specs that can find your keys, navigate you in 3D, and take agentic action — no renders, real hardware, real time Plus, Project Moohan when you want to lock in See the future of AI + XR in Shahram’s TED talk 👇 https://x.com/bilawalsidhu/status/1912884971137900631
“1/ Box + Google ADK With @Box + @Google’s Agent Development Kit (ADK), you can craft agents that handle complex tasks, like a tool for your agent to tap into Box’s capabilities and streamline your workflows like never before! https://x.com/AtomSilverman/status/1914761720914174312
“Gemini 2.5 Pro continues to make great progress in completing Pokémon! Just earned its 5th badge (next best model only has 3 so far, though with a different agent harness) 👀 https://x.com/OfficialLoganK/status/1913365614397182096
“Google Agent Development Kit – Getting Started Guide Here is my quick guide on how to get started with ADK. I presented it in our live office hours, but I think it could be useful to some of you. It’s WIP (bookmark it!) https://x.com/omarsar0/status/1915402607574893052
“Let’s build & deploy production grade Gemini AI Agents in 3 simple steps using Google Cloud Vertex AI Engine. Works with LangChain, LlamaIndex, and other Agent frameworks. https://x.com/Saboo_Shubham_/status/1911607097659416880
“Battling Google’s dominance and stronghold of what apps go into Android is a massive challenge. We had to do this through building a much better native Assistant than Gemini, and finiding an agile partner in Motorola to give consumers a great end user experience.” / X https://x.com/AravSrinivas/status/1915438656635933130
“🔥 like what I see from the Google Agent Development Kit (ADK) let’s you spin up a web ui: 1. type ‘adk web’ in the terminal 2. run the agent with the chat ui 3. view tool calls, LLM calls and responses in the side bar this is a simple example from the docs to get started https://x.com/daniel_mac8/status/1910428764435022014
“Attention @Box & @Google developers! Want to supercharge your agentic workflows with Box? Check out the sneak peek using Google’s new ADK! Agentic Workflows Unleashed: With Google’s Agent Development Kit (ADK), you can craft agents that handle complex tasks, like a tool for your https://x.com/BoxPlatform/status/1909621349015859593
“18/ Gemini AI Agent team can now help you get over your breakup. @Saboo_Shubham_ built a Breakup Recovery Agent team with Google Gemini Flash-2.0 that can analyze your chats and understand your feelings to help you with the breakup. https://x.com/AtomSilverman/status/1913372946300833837
“Use LlamaIndex to build A2A-compatible agents! A2A (Agent2Agent) is an open protocol launched by @google and supported by over 50 technology partners that enables AI agents to communicate with each other, securely exchange information, and coordinate actions, regardless of their https://x.com/llama_index/status/1912949446322852185
“I built an MCP AI Agent using Gemini Flash 2.5 with access to AirBnB and Google Maps in just 30 lines of Python Code. 100% Opensource Code. https://x.com/Saboo_Shubham_/status/1913056698635587804
“14/ @Saboo_Shubham_built an MCP AI Agent using Gemini Flash 2.5 with access to AirBnB and Google Maps in just 30 lines of Python Code. https://x.com/AtomSilverman/status/1913372938293882978
“devs using @auth0 should be able to ask: “I want to create a mobile app with passkey and Google Sign-In, and access to an API with these scopes…” and we’d just set everything up for them with the new official @auth0 MCP server, you can 💪 @cursor_ai demo 📺👇, link in reply https://x.com/dschenkelman/status/1912314395406848173
Google Leads With LLMs, Meta and OpenAI Struggle – IEEE Spectrum https://spectrum.ieee.org/large-language-models-2025
“The thing that is particularly frustrating about Google to us is how hard it is to change anything on Android. OEMs can only use a Google-approved version of Android, if they want to have core Google apps like PlayStore, Maps, etc. And “Google-approved” means keeping Google as” / X https://x.com/AravSrinivas/status/1914374321470083561
“Introducing Gemini 2.5 Flash 🔥 ⚡️High quality for best cost 🤔Dynamic thinking, with fine-grained thinking budget 🚅Available on AI Studio Blog: https://x.com/osanseviero/status/1912967087523344711
“Gemini 2.5 Flash Aider Polyglot results I guess Google does need a specialized coding model after all” / X https://x.com/scaling01/status/1914096922471932066
“Geobench – A benchmark to measure how well llms can pinpoint the location based on a Google Streetview image. Basically it makes llms play the game GeoGuessr, and find out how well each model performs on common metrics in the GeoGuessr community – if it guess the correct https://x.com/rohanpaul_ai/status/1913350223247683980
Try generating video in Gemini, powered by Veo 2 https://blog.google/products/gemini/video-generation/
Gemma 3 QAT Models: Bringing state-of-the-Art AI to consumer GPUs – Google Developers Blog https://developers.googleblog.com/en/gemma-3-quantized-aware-trained-state-of-the-art-ai-to-consumer-gpus/
“🧵 Today marks 10 years of @googlefi 🎊, which was created to bring people wireless connectivity that’s simple and secure.” / X https://x.com/Google/status/1914738922795753634
Google Gemini has 350M monthly users, reveals court hearing | TechCrunch https://techcrunch.com/2025/04/23/google-gemini-has-350m-monthly-users-reveals-court-hearing/
“gemini 2.5 is distilled from yandexGPT, it’s over” / X https://x.com/teortaxesTex/status/1914698613726326879
“The future of AI + XR starts here. Glasses that can see and think alongside you. Here’s what Google’s been cooking — its first on‑camera demo, fresh off the TED stage. https://x.com/bilawalsidhu/status/1912893463097442492
“Researchers at DeepMind released a new paper proposing ‘experiential’ AI learning rather than training based on human-generated data! The approach would allow AI to learn with extended real-world interactions and feedback, and improve over time! https://x.com/rowancheung/status/1914201357042491774
“NEW: Quantisation Aware Trained Gemma 3 from @GoogleDeepMind 🔥 > Gemma 3 27B from 54GB → 14.1GB) 🤯 Open source checkpoints – Run today in MLX, llama.cpp, lmstudio and more 💥 https://x.com/reach_vb/status/1913221589115478298
Start building with Gemini 2.5 Flash – Google Developers Blog https://developers.googleblog.com/en/start-building-with-gemini-25-flash/
“Excited to do the full release of Gemma 3 QAT! 🤏New Gemma checkpoints that can be quantized with minimal quality degradation 🤗We’re excited to have day-0 launch with @ollama, MLX, llama.cpp, @lmstudio, and Gemma.cpp Developer friendly models ftw!🚀 https://x.com/osanseviero/status/1913220285328748832
What’s next for AI at DeepMind, Google’s artificial intelligence lab | 60 Minutes – CBS News https://www.cbsnews.com/video/demis-hassabis-ai-deepmind-60-minutes-video-2025-04-20/
“We’re at #ICLR2025 in Singapore! 🇸🇬 Together with Google Research, our teams will be presenting over 125 papers, hosting workshops and showcasing demos – including Project Astra. Come say hi at our booth! 👋 https://x.com/GoogleDeepMind/status/1915216622085656688
“Gemma 3 now has torch.compile support 🔥 This means ~60% inference in @huggingface transformers in consumer GPUs after an initial warm up compilation ⚡️ Make sure to install with https://x.com/osanseviero/status/1914264677170753656
“Google is expanding Gemini Live’s Project Astra capabilities—to all Android users! This will enable users to engage with real-time visual AI and have multilingual conversations about anything seen and heard via their phone’s camera or screen share https://x.com/rowancheung/status/1912744138732154922
Improving brain models with ZAPBench https://research.google/blog/improving-brain-models-with-zapbench/
“At this point I launch 1-3 Deep Research reports anytime I want to buy something even a little significant (OpenAI tends to do a better job identifying top products and finding hidden gems, Google does better at ratings and where to buy).” / X https://x.com/emollick/status/1914516982860538312
“Breaking: new @OpenAI models shake up the Arena leaderboard🔥 Highlights: – o3 #2 overall, ties Gemini-2.5-Pro at #1 in Style Control, Math, Coding, and Hard Prompts – o4-mini breaks into top 10 and claims #1 in Math, surpassing o1 (!) – GPT-4.1 ranks top-5 in Hard Prompts, https://x.com/lmarena_ai/status/1915078057452573142
“today’s google doodle (for earth day) goes hard af https://x.com/bilawalsidhu/status/1914829248679788758
“Google introduced Gemini 2.5 Pro Experimental, the first in its new Gemini 2.5 family, designed with enhanced reasoning and coding abilities. Trained with reinforcement learning to generate hidden reasoning steps, the model currently tops the Chatbot Arena leaderboard and https://x.com/DeepLearningAI/status/1913019660880032169
“Google dropped Gemini 2.5 Flash, a reasoning model matching o4-mini in preview It’s ‘thinking budget’ (up 24k tokens), can balance between answer quality, cost, and speed The model performs particularly well on reasoning, STEM, and visual reasoning https://x.com/adcock_brett/status/1913986859148996798
The Era of Experience Paper.pdf https://storage.googleapis.com/deepmind-media/Era-of-Experience%20/The%20Era%20of%20Experience%20Paper.pdf
Image segmentation using Gemini 2.5 https://simonwillison.net/2025/Apr/18/gemini-image-segmentation/
2025 Q1 Earnings Call – Alphabet Investor Relations https://abc.xyz/2025-q1-earnings-call/
“Why OpenAI doesn’t include other models in their own benchmarks: OpenAI-MRCR results with Gemini 2.5 Pro https://x.com/scaling01/status/1913955228442833032
Introducing Mobility AI: Advancing urban transportation https://research.google/blog/introducing-mobility-ai-advancing-urban-transportation/
Let’s finish image
Music AI Sandbox, now with new features and broader access – Google DeepMind https://deepmind.google/discover/blog/music-ai-sandbox-now-with-new-features-and-broader-access/
“I am frequently asked: “when will AI work with well spreadsheets?” Quietly, Google has come some way in realizing this vision, using AI at both the cell & sheet level (and running code). Here are examples of me taking fake startup financials I use in class and it spotting issues https://x.com/emollick/status/1913334200847265986
“.@GoogleAI has dropped a very interesting study They introduced new types of attentional bias strategies in LLMs and reimagined the “forgetting” process, replacing it with “retention.” All of this is wrapped up in Miras – their new framework for designing efficient AI https://x.com/TheTuringPost/status/1914316647386714289
“The remedy that is right in our opinion is not a breakup of Google; but rather offering consumers the choice to pick their defaults on Android without feeling the risk of a loss in revenue. That’s what we will be proposing. Read more here: https://x.com/AravSrinivas/status/1914374839722500444
Google’s surprise Android XR glasses tipped to land in 2026 – and my only complaint is they aren’t launching sooner | TechRadar https://www.techradar.com/computing/virtual-reality-augmented-reality/googles-surprise-android-xr-glasses-tipped-to-land-in-2026-and-my-only-complaint-is-they-arent-launching-sooner
Waymo reports 250,000 paid robotaxi rides per week in U.S. https://www.cnbc.com/2025/04/24/waymo-reports-250000-paid-robotaxi-rides-per-week-in-us.html
“Human generated data has fueled incredible AI progress, but what comes next? 📈 On the latest episode of our podcast, @FryRsquared and David Silver, VP of Reinforcement Learning, talk about how we could move from the era of relying on human data to one where AI could learn for https://x.com/GoogleDeepMind/status/1910363683215008227
“”In February of 2025, Americans used their television sets to watch YouTube more than anything else. More than Disney. More than HBO. Even more than Netflix.” The most underestimated factor when trying to understand the future of media is the fact that billions of people out https://x.com/c_valenzuelab/status/1915514295816737140
