Meta has released the Llama 4 models—Scout and Maverick—featuring native multimodality, mixture-of-experts architecture, and state-of-the-art context lengths. Meta also previewed Behemoth, a trillion-parameter model that supposedly outperforms GPT-4.5 and Claude 3.7 Sonnet on STEM benchmarks.
Google Cloud Next '25 showcased powerful new hardware like Ironwood TPUs, advanced models across all modalities, a comprehensive multi-agent ecosystem, and purpose-built AI agents delivering tangible business results across customer service, creative work, data analysis, coding, and security.
OpenAI has secured a record-breaking $40 billion funding round led by SoftBank, valuing the company at $300 billion, with plans to advance AI research and scale infrastructure as ChatGPT reaches 500 million weekly users.
AI drug design company Isomorphic Labs has secured $600 million in funding led by Thrive Capital to accelerate development of its AI platform, expand partnerships with pharmaceutical giants, and advance therapeutic programs into clinical trials.
Runway has released Gen-4, its most advanced AI video generation model yet, shortly before announcing it secured $308 million in Series D funding, positioning the company as a major player in the evolving AI media creation landscape despite legal challenges and industry disruption concerns.
Amazon unveiled Nova Act, an AI model for browser-based task completion. Available as a research preview, Nova Act prioritizes reliability in executing simple commands rather than higher-level workflows as the key to unlock genuine AI agents that are both capable and autonomous.
Microsoft launched two AI reasoning agents for 365 Copilot. Researcher handles complex research using multiple sources, while Analyst functions as a data scientist to transform raw data into insights. The agents will roll out this month as part of a new program called "Frontier".
Google has launched Gemini 2.5 Pro, its most intelligent "thinking model" to date. The model leads in reasoning and coding benchmarks, and features a 1M token context window. Gemini 2.5 Pro is now available for Gemini Advanced users, with Vertex AI integration coming soon.
Browser Use has raised $17 million in seed funding, led by Felicis Ventures, to build infrastructure that converts website interfaces into structured text. This will enable AI agents to navigate the web more effectively and reliably than vision-based approaches.
The ARC Prize Foundation has officially released the ARC-AGI-2 to challenge current foundation models and help track progress towards AGI. Additionally, the Foundation has opened the ARC Prize 2025, running from Mar 26 to Nov 3, with a $700K Grand Prize for an 85% scoring solution on the ARC-AGI-2.
Hugging Face has released HuggingSnap, an iOS app that runs SmolVLM2 locally and efficiently to enable visual understanding and text-based generative AI tasks, such as answering questions about an image or identifying and describing objects from visual input.
Highlights from the Nvidia GTC 2025 keynote, OpenAI's new tools for building agents, Gleamer's recent acquisitions, and OpenAI's decision to invest in CoreWeave as tensions with Microsoft continue to rise.