By the TensorMax editorial team
· Drawing from sources across the AI industry
Today's top story
model release
The Google Antigravity SDK is a Python SDK for building AI agents powered by Antigravity and Gemini.
Why it matters. The release of the Google Antigravity SDK is a significant development in the AI industry, providing a Python library for building AI agents powered by Antigravity and Gemini. With over 10 examples and a three-layer architecture, this SDK offers a secure, scalable, and stateful infrastructure layer that abstracts the agentic loop, allowing developers to focus on what their agent does rather than how it runs. The SDK's capabilities, including support for rich multimedia file attachments and declarative policy systems, make it an attractive option for developers looking to build complex AI agents.
The Google Antigravity SDK is a Python SDK for building AI agents powered by Antigravity and Gemini. It provides a secure, scalable, and stateful infrastructure layer that abstracts the agentic loop, letting developers focus on what their agent does rather than how it runs. The SDK can be installed using pip install google-antigravity and provides a range of features, including support for rich multimedia file attachments, declarative policy systems, and background tasks that react to external events. The SDK follows a three-layer architecture, consisting of the Agent, Connections, Conversation, Hooks, MCP, Tools, and Triggers components. Developers can use the SDK to build complex AI agents, with features such as streaming agent output in real-time, registering Python functions as tools, and connecting to external MCP servers. The SDK also provides a range of examples, including a hello_world example, to help developers get started. With its robust feature set and flexible architecture, the Google Antigravity SDK is poised to have a significant impact on the AI industry. The SDK's documentation provides more detailed information on specific components, including the Agent, Connections, Conversation, Hooks, MCP, Tools, and Triggers. Overall, the Google Antigravity SDK is a powerful tool for building AI agents, and its release is a significant development in the AI industry.
More from today
funding event
$100B
Why it matters. The strategic stake of this signal lies in the fact that addresses linked to Iran, Russia, North Korea, and other US-sanctioned entities received over $100 billion in crypto in 2025, which is almost 8 times the amount in 2024. This significant increase in crypto transactions among blacklisted entities has major implications for global security and the use of cryptocurrency in financing terrorism and weapons.
funding event
$20B
Why it matters. The $20 billion licensing and talent deal between Nvidia and Groq has significant implications for the AI chip market, with Nvidia gaining key talent, including Groq's former CEO Jonathan Ross, and access to Groq's custom AI inference chips. This deal allows Groq to remain independent while Nvidia strengthens its position in the market. The acquisition of Ross and other key engineers brings valuable expertise to Nvidia, potentially accelerating its development of AI technologies. With a deal of this magnitude, Nvidia is poised to further solidify its leadership in the AI chip industry, putting pressure on competitors to respond.
enterprise adoption
Why it matters. The strategic stake of Elbit Systems' Tzayad digital army program is evident in its ability to identify 850,000 targets in Gaza and Lebanon wars, with the Israeli military detecting about 1,000 potential targets a day during the first two years of the wars. This capability has significant implications for the future of warfare, particularly in terms of the use of artificial intelligence to support tactical decision-making. The program's effectiveness in mapping the positions of friendly units and enemy targets has allowed the Israeli military to increase the speed of external fire support, reducing the time it takes to launch attacks from 40 to 50 minutes to just one to seven minutes.
regulatory action
Why it matters. The US government's restriction of Anthropic's Mythos class of models, citing supply chain risk, has significantly broken down trust in closed-source models and companies like Anthropic and OpenAI. This shift has led to a marked change in the future of AI, with a growing adoption of open-weight models out of China. The restriction, which occurred on June 12th, 2026, has also led to a division in the AI community over closed source vs. open-source models, with companies like Palantir openly questioning the value proposition of closed-source model makers. According to the Pentagon, Anthropic was formally designated as a supply chain risk on February 27, 2026, and the Trump Administration's directives are seeking to have power over how America's AI leaders release their models, further accelerating the adoption of China's open-weight models.
funding event
Why it matters. The Trump Administration's recent actions have significantly damaged America's AI leadership, with the restriction of Anthropic's Mythos class of models and Palantir's argument for open-weight models favoring China. This has led to a big transfer from closed-source models to open-source models, with many businesses pivoting to Chinese open-weight models due to their cost-effectiveness. Sam Altman's offer of a 5% government stake in OpenAI to the Trump Administration has also raised concerns about the US government's involvement in picking winners in AI, with potential geopolitical and reputational damage for the AI industry. According to JPMorgan research, total LLM token volume surged by 70% month-over-month from May to June, indicating a shift towards more efficient models.
market event
Why it matters. The strategic stake of the Token Apocalypse is that it marks a significant shift in the AI industry, with companies pivoting from closed-source models to open-weight models, driven by the 70% month-over-month surge in token volume. This shift is largely due to the restricted access to Anthropic's Mythos class of models, which has led to a decline in trust in closed-source models and companies like Anthropic and OpenAI. As a result, open-weight models from China are gaining traction, with companies like Palantir and Nvidia partnering to deliver secure 'Sovereign AI' solutions. This trend has the potential to slow down the growth of AI behemoths and model makers, and could have a material impact on the AI bubble.
Catch up quick
-
Nvidia delays its next-gen AI rack system Kyber NVL144 by 12+ months to 2028
-
AI tools from Meta, Google, Alibaba, and Mistral introduce bias into user drafts on sensitive topics like abortion and climate change, potentially reshaping public opinion
-
JADEPUFFER, an AI-powered ransomware operation, exploited a known vulnerability in Langflow to steal credentials and destroy databases
-
Japan to invest up to $6.2 billion in SoftBank-led AI models
-
Harvard Business School and INSEAD study finds AI-native startups hire fewer juniors and more elite-educated workers
-
Tesla launches Robotaxi service in Miami without a safety monitor
-
Huawei's upcoming Kirin 2026 mobile processor to power Mate handsets with 55% increased transistor density
-
Comcast's Sky buys UK's ITV for £2.1 billion
-
Apple releases PathMoE, a new Mixture-of-Experts architecture for natural language processing
-
Alejandro Tlaie probes the loss-band sparsity assumption in Scientist AI, finding the dangerous set has negligible local volume but the loss landscape doesn't actively protect that sparsity
-
Qwen 72B model's system prompt adherence degrades from 98% to 48% over 50 conversation turns
-
Semiconductor stocks post biggest quarterly gain on record, driven by AI demand
-
China's EV deliveries fell 7% year-over-year in June, with 1.04 million units sold, exacerbating industry worries about profitability
-
Apple Machine Learning Research publishes paper on Scaling Properties of Continuous Diffusion Spoken Language Models at Interspeech 2026
-
Honeywell spinoff Solstice Advanced Materials and Element Solutions in talks to merge
Also on the desk
Who else should be reading this?Hit reply with the name and email of one person you think belongs on this list. We'll reach out personally, with your intro if you want.
|
|