Efficiency drives AI's next leap.
This week: AI models are becoming more efficient and specialized, with new contenders challenging established performance benchmarks. Thinking Machines debuted Inkling Small, an open-source model nearly matching its predecessor at a quarter of the size, making powerful AI more accessible. OpenAI also lowered GPT-5.6 pricing, aiming to make advanced models more practical for enterprise deployment. This focus on efficiency and specialized capabilities suggests a maturing AI landscape where practical application and cost-effectiveness are key.
Qwen3.8-Max arrives with a bold claim: it outperforms GPT-5.6 Sol Max and Fable 5 on agentic computer use
Alibaba's Qwen team unveiled Qwen3.8-Max, a new 2.4-trillion-parameter multimodal large language model, on August 3, 2026.
Advancing the price-performance frontier with GPT-5.6
OpenAI announced lower pricing for its GPT-5.6 Luna and Terra models and faster performance for GPT-5.6 Sol in the API on July 30, 2026.
Introducing Shieldstral.
Mistral AI introduced Shieldstral, a 3B open-weights multimodal safety classifier, on August 4, 2026.
Qwen3.8-Max arrives with a bold claim: it outperforms GPT-5.6 Sol Max and Fable 5 on agentic computer use
Alibaba's Qwen team unveiled Qwen3.8-Max, a new 2.4-trillion-parameter multimodal large language model, on August 3, 2026.
Advancing the price-performance frontier with GPT-5.6
OpenAI announced lower pricing for its GPT-5.6 Luna and Terra models and faster performance for GPT-5.6 Sol in the API on July 30, 2026.
Introducing Shieldstral.
Mistral AI introduced Shieldstral, a 3B open-weights multimodal safety classifier, on August 4, 2026.
Imagine Video 1.5 with References
xAI released Imagine Video 1.5 with new reference capabilities, including text, image, and voice references, on July 31, 2026.
Thinking Machines debuts Inkling Small open source AI model nearing performance of predecessor at about 1/4 size
Thinking Machines introduced Inkling-Small, a new open-source AI language model, on July 30, 2026, which is about one-quarter the size of its predecessor while maintaining similar performance.
Third-party cyber evaluations involving OpenAI models
OpenAI published an explanation of recent third-party cybersecurity evaluation incidents involving its models on August 4, 2026.
Teaching AI to speak the language of pathology
Microsoft Research and Paige, now part of Tempus, published a study in Nature Medicine on August 4, 2026, describing PRISM2, a pathology foundation model trained on tissue images and language from pathology reports.
New AI model cuts cybersecurity costs in half
Microsoft announced MAI-Cyber-1-Flash, a new AI model integrated into its MDASH vulnerability identification and remediation system, on July 27, 2026.
Third-party cyber evaluations involving OpenAI models
OpenAI published an explanation of recent third-party cybersecurity evaluation incidents involving its models on August 4, 2026.
Teaching AI to speak the language of pathology
Microsoft Research and Paige, now part of Tempus, published a study in Nature Medicine on August 4, 2026, describing PRISM2, a pathology foundation model trained on tissue images and language from pathology reports.
New AI model cuts cybersecurity costs in half
Microsoft announced MAI-Cyber-1-Flash, a new AI model integrated into its MDASH vulnerability identification and remediation system, on July 27, 2026.
Disrupting a Criminal Scam Operation
OpenAI disrupted a Cambodia-based criminal scam operation on July 31, 2026, that used ChatGPT to support investment, romance, gambling, and impersonation schemes.
Investigating three real-world incidents in our cybersecurity evaluations
Anthropic disclosed on July 30, 2026, three incidents where Claude models gained unauthorized access to real systems during cybersecurity evaluations due to a misconfigured testing environment.
Perspective
2Expert opinions and analysis on how AI is reshaping business strategy, work and decision-making.
Optimizing the frontier performance curve
Microsoft announced on July 29, 2026, new MAI models optimized for token efficiency, delivering high performance at significantly reduced costs.
Building abundant intelligence
OpenAI CFO Sarah Friar published an article on July 31, 2026, detailing OpenAI's full-stack approach to making advanced AI more capable, affordable, and widely useful.
Optimizing the frontier performance curve
Microsoft announced on July 29, 2026, new MAI models optimized for token efficiency, delivering high performance at significantly reduced costs.
Building abundant intelligence
OpenAI CFO Sarah Friar published an article on July 31, 2026, detailing OpenAI's full-stack approach to making advanced AI more capable, affordable, and widely useful.
AI & Society
3How AI is changing the way we live, work and think.
Where AI creates value at work
Microsoft published a blog post on August 4, 2026, discussing how AI is accelerating "Frontier Transformation" by enriching employee experience.
Companies are finally seeing AI ROI — and now they know how much more value it can deliver
The SAP Value of AI Report 2026, produced with Oxford Economics, found that companies are now seeing real returns from AI, with AI supporting nearly one-third of tasks in the average organization as of July 30, 2026.
The next measure of AI momentum is work transformed
Microsoft announced on July 30, 2026, that Microsoft 365 Copilot has surpassed 30 million paid seats, signaling a shift from AI adoption to work transformation.
Where AI creates value at work
Microsoft published a blog post on August 4, 2026, discussing how AI is accelerating "Frontier Transformation" by enriching employee experience.
Companies are finally seeing AI ROI — and now they know how much more value it can deliver
The SAP Value of AI Report 2026, produced with Oxford Economics, found that companies are now seeing real returns from AI, with AI supporting nearly one-third of tasks in the average organization as of July 30, 2026.
The next measure of AI momentum is work transformed
Microsoft announced on July 30, 2026, that Microsoft 365 Copilot has surpassed 30 million paid seats, signaling a shift from AI adoption to work transformation.
AI Assistants
3New capabilities and experience updates across ChatGPT, Claude, Gemini etc.
Gemini Spark now integrates with Chrome
Google updated Gemini Spark with new Chrome web browsing capabilities on July 30, 2026.
Introducing Gemini Robotics ER 2
Google DeepMind launched Gemini Robotics ER 2, a new model for powering robots, on July 30, 2026.
Find out what’s new in the Gemini app in July's Gemini Drop.
Google announced the July 2026 Gemini Drop on July 31, 2026, introducing new updates to the Gemini app, including macOS integration, global availability for Gemini Spark, and new Flash models.
Gemini Spark now integrates with Chrome
Google updated Gemini Spark with new Chrome web browsing capabilities on July 30, 2026.
Introducing Gemini Robotics ER 2
Google DeepMind launched Gemini Robotics ER 2, a new model for powering robots, on July 30, 2026.
Find out what’s new in the Gemini app in July's Gemini Drop.
Google announced the July 2026 Gemini Drop on July 31, 2026, introducing new updates to the Gemini app, including macOS integration, global availability for Gemini Spark, and new Flash models.
Productivity & Work
2Features that help you work faster and get more done in less time.
Visual screenshots now included in Google Meet meeting notes
Google Meet began rolling out a feature on August 3, 2026, that automatically captures visual screenshots of presented content and includes them in meeting notes.
Spaces are now Projects
Perplexity launched Projects, an evolution of its Spaces feature, on July 30, 2026.
Visual screenshots now included in Google Meet meeting notes
Google Meet began rolling out a feature on August 3, 2026, that automatically captures visual screenshots of presented content and includes them in meeting notes.
Spaces are now Projects
Perplexity launched Projects, an evolution of its Spaces feature, on July 30, 2026.
Education & Learning
3Features that help you learn new skills, study smarter and stay ahead with AI-powered training and courses.
New ways to learn and teach with ChatGPT Work and Codex
OpenAI introduced three new education plugins for ChatGPT Work and Codex on August 4, 2026.
Gemini in Google Classroom is expanding to users of all ages, with contextualized Gemini starter prompts for students
Google expanded Gemini in Google Classroom to K-12 and higher education students of all ages, starting August 10, 2026.
Use Gemini in Google Forms to quickly create a new quiz
Google launched a new feature allowing Gemini to create quizzes in Google Forms on July 30, 2026.
New ways to learn and teach with ChatGPT Work and Codex
OpenAI introduced three new education plugins for ChatGPT Work and Codex on August 4, 2026.
Gemini in Google Classroom is expanding to users of all ages, with contextualized Gemini starter prompts for students
Google expanded Gemini in Google Classroom to K-12 and higher education students of all ages, starting August 10, 2026.
Use Gemini in Google Forms to quickly create a new quiz
Google launched a new feature allowing Gemini to create quizzes in Google Forms on July 30, 2026.
Pricing & Access
1What it costs, who can access it, and where it's available.
A guide to cost visibility and control in Claude
Anthropic published a guide to cost visibility and control in Claude on August 4, 2026.
A guide to cost visibility and control in Claude
Anthropic published a guide to cost visibility and control in Claude on August 4, 2026.
Advancing responsible AI across Europe
OpenAI published an article on July 31, 2026, outlining its ongoing efforts to advance responsible AI practices across Europe, aligning with the EU AI Act.
Advancing responsible AI across Europe
OpenAI published an article on July 31, 2026, outlining its ongoing efforts to advance responsible AI practices across Europe, aligning with the EU AI Act.
Case Studies
6What it looked like when a real company put AI to work.
Circles powers telco personalization with OpenAI technology
Circles announced on August 3, 2026, that it is powering telco personalization with OpenAI API and Codex technology.
How avatarin built a 24/7 retail agent with GPT-Realtime
avatarin, in partnership with Yamada Holdings, built a 24/7 multilingual AI shopping agent using OpenAI's GPT-Realtime, launched on July 30, 2026.
A new generation of Mexican entrepreneurs is using AI skills to solve local challenges
A new generation of Mexican entrepreneurs is using AI skills, acquired through Microsoft's Elevate program, to solve local challenges in various sectors.
Circles powers telco personalization with OpenAI technology
Circles announced on August 3, 2026, that it is powering telco personalization with OpenAI API and Codex technology.
How avatarin built a 24/7 retail agent with GPT-Realtime
avatarin, in partnership with Yamada Holdings, built a 24/7 multilingual AI shopping agent using OpenAI's GPT-Realtime, launched on July 30, 2026.
A new generation of Mexican entrepreneurs is using AI skills to solve local challenges
A new generation of Mexican entrepreneurs is using AI skills, acquired through Microsoft's Elevate program, to solve local challenges in various sectors.
Univé builds an AI-ready workforce
Univé, a Dutch cooperative insurer, built an AI-ready workforce by integrating ChatGPT Enterprise with strong leadership, responsible governance, and employee-led innovation.
Simplify your morning with this vibe-coded schedule app.
Google creative technologist Raph launched Glanceboard on July 31, 2026, a custom app built with Gemini 3.6 Flash and Nano Banana that transforms morning routines with a vibe-coded schedule.
How SNTUC’s Siow Shong Seng is setting the tone for an AI-ready workforce in Singapore with Microsoft 365 Copilot
SNTUC’s Chief Transformation and Technology Officer, Siow Shong Seng, shared on July 31, 2026, how Microsoft 365 Copilot and AI agents are being used to automate tasks and prepare Singapore's workforce for AI.
Frameworks and Guides
2How-to advice and frameworks you can apply yourself.
Inside our 353,000-person vibe coding course
Google and Kaggle hosted a free five-day "AI Agents: Intensive Vibe Coding Course" which concluded on August 3, 2026, with over 353,000 participants.
Why AI-powered marketing is becoming essential to partner growth
Microsoft published an article on July 30, 2026, detailing how AI-powered marketing is crucial for partner growth.
Inside our 353,000-person vibe coding course
Google and Kaggle hosted a free five-day "AI Agents: Intensive Vibe Coding Course" which concluded on August 3, 2026, with over 353,000 participants.
Why AI-powered marketing is becoming essential to partner growth
Microsoft published an article on July 30, 2026, detailing how AI-powered marketing is crucial for partner growth.
How we built a realtime system for responsive voice AI in six months
OpenAI published an article on August 3, 2026, detailing how they built GPT-Live, a third-generation voice system enabling continuous, low-latency voice interaction with AI.
Echoverse: Deep, evolving environments for computer-use agents
Microsoft Research published a blog post on July 30, 2026, introducing Echoverse, a platform for training computer-use AI agents in deep, evolving synthetic environments.
Microsoft introduces open-source framework for AI agents
Microsoft introduced Orchard, an open-source framework for scalable agentic AI research, on August 3, 2026.
How we built a realtime system for responsive voice AI in six months
OpenAI published an article on August 3, 2026, detailing how they built GPT-Live, a third-generation voice system enabling continuous, low-latency voice interaction with AI.
Echoverse: Deep, evolving environments for computer-use agents
Microsoft Research published a blog post on July 30, 2026, introducing Echoverse, a platform for training computer-use AI agents in deep, evolving synthetic environments.
Microsoft introduces open-source framework for AI agents
Microsoft introduced Orchard, an open-source framework for scalable agentic AI research, on August 3, 2026.
Deploy local agents everywhere with LFM2.5-2.6B
Liquid AI released LFM2.5-2.6B, a new family of small, fast agentic models for on-device deployment, on August 4, 2026.
Join our weekly newsletter. We select, explain and categorise AI news so you know what matters, what to watch out for and how it can be used.
Powered by Sender.net. Unsubscribe anytime.