Latest News6 / 7
AI tool launches, product updates and industry moves.
The White House's 'Winning the Race': America's AI Action Plan Deregulates on 90 Fronts
On July 23, 2025, the Trump administration released its AI Action Plan and signed three executive orders: accelerate innovation, build infrastructure, export the American AI stack.
AI Wins Its First Math Olympiad Gold: Google and OpenAI Both Reach the Standard
In July 2025, Google DeepMind's Gemini Deep Think and OpenAI's experimental model both hit gold-medal level at the IMO — LLMs matching gold medalists for the first time.
Windsurf Changes Hands Three Times in 72 Hours: OpenAI Out, Google Pays $2.4B, Cognition Takes the Rest
In July 2025 Windsurf lived the wildest 72 hours in startup history: after the OpenAI deal collapsed, Google paid $2.4B for the founders and Cognition bought whatever was left.
Moonshot AI Releases Kimi K2: A Trillion-Parameter MoE Model, Open-Sourced
Moonshot AI unveils Kimi K2 with 1 trillion total parameters — the largest open model at launch — excelling at agentic and coding tasks.
Grok 4 Launches: xAI's 200,000-GPU Bid for the Benchmark Crown, With a $300/mo Heavy Tier
xAI released Grok 4 in July 2025, setting records on GPQA and other hard benchmarks, plus a multi-agent Heavy edition.
Waymo Passes 100 Million Driverless Miles as Robotaxis Reach Real Scale
In 2025 Waymo logged 100 million fully autonomous miles and tripled annual rides to 15 million — robotaxis went truly commercial.
Meta Pours $14.3B Into Scale AI as 28-Year-Old Alexandr Wang Takes Over Superintelligence Labs
In June 2025 Meta bought 49% of Scale AI for $14.3 billion and recruited its CEO — the most expensive shot in the AI talent war.
Anthropic Releases Claude 4: Opus 4 and Sonnet 4 Debut Together
At its first Code with Claude developer conference, Anthropic unveiled Claude Opus 4 and Sonnet 4, built for hours-long autonomous coding.
GitHub Copilot Gets a Coding Agent: Autonomous Background Tasks and Pull Requests
At Microsoft Build 2025, GitHub introduced the Copilot coding agent — it takes tasks, runs in the background on GitHub Actions and opens PRs.
AlphaEvolve: DeepMind's AI Breaks a 56-Year Matrix Multiplication Record
In May 2025, DeepMind unveiled the Gemini-powered coding agent AlphaEvolve, which found a 4x4 matrix multiplication method beating Strassen's algorithm.
Qwen 3 Released: MoE Architecture Sets a New Bar for Open Source
Alibaba releases the Qwen 3 series; the flagship Qwen3-235B-A22B rivals top models like DeepSeek-R1 and o1 on major benchmarks.
OpenAI Releases o3 and o4-mini: Reasoning Models Gain Full Tool Access
OpenAI launches o3 and o4-mini, the first reasoning models to integrate web browsing, Python, and image analysis directly into their chain of thought.
$32 Billion With No Product: Ilya Sutskever's SSI Raises Another $2B
In April 2025, Safe Superintelligence raised $2 billion at a $32 billion valuation — under a year old, zero product, zero revenue.
Meta Releases Llama 4: Scout and Maverick Debut as Natively Multimodal Open Models
Meta released Llama 4 Scout and Maverick on April 5, 2025 — its first natively multimodal open-weight models, with Scout supporting a 10M-token context.
Devin 2.0 Released: Price Drops from $500 to $20 a Month
Cognition releases Devin 2.0 with a collaborative IDE and a starting price slashed from $500 to $20 per month, taking the AI software engineer mainstream.
Midjourney V7 Released: First All-New Image Model in Nearly a Year
Midjourney launched V7 on April 3, 2025, with clear gains in prompt understanding, image quality and hand details, plus default personalization.
Runway Releases Gen-4: Full-Film Character Consistency from One Reference Image
Runway released Gen-4 on March 31, 2025, built on world consistency: a single reference image keeps characters stable across scenes.
xAI Acquires X in an All-Stock Deal: Musk Folds the Social Platform Into His AI Company
On March 28, 2025, Musk announced xAI's all-stock acquisition of X — valuing xAI at $80 billion and X at $33 billion — merging data and distribution.
Anthropic Gives LLMs a 'Brain Scan': An Interpretability Breakthrough in Tracing Thoughts
In March 2025, Anthropic published interpretability research that first 'traces' Claude's internal thought circuits, revealing how LLMs think and why they hallucinate.
NYT v. OpenAI Cleared to Proceed: The Copyright Case of the Century Heads to the Merits
On March 26, 2025, a federal judge rejected OpenAI's motion to dismiss, letting The New York Times's core copyright claims proceed — with Microsoft as co-defendant.
Google Releases Gemini 2.5 Pro: Its First Thinking Flagship Tops the Benchmarks
Google released Gemini 2.5 Pro in March 2025 with built-in reasoning, leading multiple benchmarks by meaningful margins.
NVIDIA GTC 2025: Jensen Huang Unveils Blackwell Ultra and the Vera Rubin Roadmap
At GTC 2025, Jensen Huang unveiled Blackwell Ultra (shipping H2 2025) and previewed the next-gen Vera Rubin architecture.
Manus Goes Viral Overnight: The General AI Agent Whose Invite Codes Sold for Thousands
Butterfly Effect launched the general-purpose agent Manus on March 6, 2025; the invite-only rollout sparked a code-hunting frenzy with a 2M+ waitlist.
Alexa+ Arrives: Amazon Gives Its 600-Million-Device Assistant a Generative Brain
On February 26, 2025, Amazon unveiled the generative-AI Alexa+, free for Prime members — the biggest overhaul in the assistant's decade.