| |
Good morning, {{ first_name | AI enthusiasts }}. Just weeks after its powerful Gemini I/O update, Google is again strengthening its grip on the AI leaderboards.
With a new 2.5 Pro update that shows improvements across the board on its already No.1 ranked predecessor, the tech giant’s rapid launches are setting a relentless new pace for AI model releases.
Reminder: Our next live workshop is today at 4 PM EST! Come learn how to incorporate the new Flux Kontext editing model into your creative workflows for consistent, high-quality results. .
In today’s AI rundown:
-
Google’s big Gemini 2.5 Pro update
-
Anthropic’s Claude Gov for U.S. agencies
-
How to translate any video into multiple languages
-
AI foot scanner predicts heart failure weeks early
-
4 new AI tools & 4 job opportunities
LATEST DEVELOPMENTS
🚀
Image source: Google
The Rundown: Google just a new update preview to its Gemini 2.5 Pro model, calling it the company’s “most intelligent model yet” — with notable jumps on coding, STEM, reasoning, and image understanding benchmarks.
The details:
-
The new model shows major performance gains, its lead on user-preference leaderboards like LMArena and WebDevArena.
-
Google specifically addressed user feedback on the previous version to fix performance regressions in non-coding tasks like creative writing.
-
The update also brings “thinking budgets” in the API to manage cost and latency, with the preview set to become an official release in the coming weeks.
-
The upgraded preview is accessible to devs via the Gemini API in AI Studio and Vertex AI, while also being deployed to the public-facing Gemini app.
Why it matters: Just weeks after its coding-enhanced 2.5 Pro I/O update, the tech giant is at it again with another upgrade that brings broader quality upgrades across the board. Google extends its lead on the leaderboards, and is also changing the release dynamic — opting for frequent ‘preview’ drops before a full model launch.
TOGETHER WITH INNOVATING WITH AI
🚀
The Rundown: Innovating With AI’s ‘The AI Consultancy Project’ gives you the frameworks, playbooks, and client‑ready templates needed to turn “interesting AI ideas” into a revenue‑generating business – helping you ride the wave of an AI consulting boom expected to grow by 8x this decade.
In this program, you will:
-
Gain the tools and frameworks to find clients and deliver top-notch services
-
Follow a 6-month plan to build a 6-figure AI consulting business
-
Join a 700‑strong cohort where some members landed their first AI client in 72 hours
.
ANTHROPIC
🏛️
Image source: Anthropic
The Rundown: Anthropic Claude Gov, a specialized version of its AI models designed exclusively for U.S. defense and intelligence agencies — featuring modified safety guardrails and enhanced capabilities for handling classified information.
The details:
-
Anthropic said the models are already deployed at the highest levels of U.S. national security, exclusively for those who handle classified information.
-
The models feature reduced refusal rates when processing classified materials and improved comprehension of defense and intelligence documentation.
-
Key enhancements target mission-critical needs, including foreign language analysis and cybersecurity pattern recognition for intelligence work.
-
The company created exemptions for government contracts while preserving restrictions on weapons design, disinformation, and malicious cyber operations.
Why it matters: With Claude Gov, ChatGPT Gov, and other AI leaders exploring tailored models for military and intelligence contracts, the tech is becoming infused within the deepest layers of the government. But the major AI labs are now walking a tightrope between their ethical and safety concerns and commercial opportunities.
AI TRAINING
🌍
The Rundown: In this tutorial, you will learn how to use HeyGen’s AI video translation feature to automatically translate your videos into different languages while preserving the original speaker’s voice and lip-sync.
Step-by-step:
-
Go to HeyGen’s and select “Translate a Video” from the dashboard
-
Upload your video file or paste a YouTube/Google Drive URL
-
Choose your source and target languages, then click “Advanced” to configure settings like dynamic duration, captions, voice enhancement, and background music removal
-
Click “Generate” to obtain your multilingual video
Pro tip: Use videos with clear speech and good lighting for the most accurate lip-sync and voice matching results.
PRESENTED BY BASETEN
⚙️
The Rundown: Baseten just launched Model APIs, featuring on-demand frontier models like DeepSeek V3, R1, and Llama Maverick, purpose-built for production workloads — running on their optimized inference stack that’s dominating performance benchmarks.
Key advantages include:
-
Built-in tool calling and structured output support for agent use cases
-
Full OpenAI compatibility means Model APIs can be drop-in replacements for closed models
-
Regular rollouts of more leading models pre-optimized for popular GenAI use cases
on pre-optimized frontier models.
AI RESEARCH
🦶
Image source: Heartfelt Technologies
The Rundown: Cambridge startup Heartfelt Technologies an AI-powered wall-mounted scanner that monitors ankle swelling, accurately predicting heart failure up to 13 days before patients need emergency care.
The details:
-
The scanner captures 1,800 images per minute of patients’ feet and ankles, using AI to measure fluid accumulation that signals worsening heart conditions.
-
In trials across five NHS trusts with 26 patients, the system predicted five out of six hospitalizations with an average warning time of 13 days.
-
The device operates automatically without requiring patient interaction, and over 80% of trial participants chose to keep the scanner after the study ended.
Why it matters: The future of healthcare is proactive, non-invasive monitoring through smart devices like this — helping identify symptoms long before a major medical event might occur and giving patients more control over their health at home instead of costly hospital stays.
QUICK HITS
🛠️
-
📊 – Now with meeting recording and data connectors
-
⚙️ – Enterprise-grade coding assistant for development tasks
-
🎥 – Luma Labs’ new tool for restyling videos
-
🎶 – Upgraded song editor, creative sliders, and extended song uploads
💼
-
📈 – Senior or Principal Product Manager, Growth
-
📊 – Data Scientist (Technical Leadership)
-
🧪 – Chromium Engineer, Android
-
🏎️ – Software Engineer, ML Accelerator
📰
ElevenLabs Eleven v3, a new text-to-speech preview model featuring emotional audio tags, multi-speaker dialogue, and support for 70+ languages.
Anthropic CEO Dario Amodei a NYT opinion piece arguing against President Trump’s ‘Big Beautiful Bill’ that would restrict state-level AI regulation for 10 years.
OpenAI the disruption of 10 malicious operations (four tied to China) that utilized ChatGPT for tasks like social media manipulation, espionage, and scams.
X its developer terms to ban the use of its content or API for AI model training, aiming to shield the social media network’s data from xAI rivals.
Bland Bland TTS, a new voice AI with enhanced realism and control for voice cloning, voice apps, and AI-powered customer support.
Volvo is a new AI-powered seatbelt, which accounts for a passenger’s size, seating position, and vehicle speed and direction to customize protection.
COMMUNITY
🎥
Join our next workshop today, June 6th at 4 PM EST with Dr. Alvaro Cintas, The Rundown’s AI professor. By the end of the workshop, you’ll confidently be able to incorporate Flux Kontext into your creative workflows for consistent, high-quality results.
RSVP . Not a member? Join on a 14-day free trial.
See you soon,
Rowan, Joey, Zach, Alvaro, and Jason—The Rundown’s editorial team
