Google's AI strategy moves fast, and Gemini sits at the center of it. Here you will find news on new Gemini models, changes to free and paid access, Google DeepMind research and how Gemini is built into Search, Workspace and Android. Each article focuses on practical impact: what is new, who gets access and how Google's approach compares with OpenAI, Anthropic and Meta.
Gemini 4 Argon reshapes Googleโs AI plans, moving free users to Flash, adding Deep Think for Pro and Ultra, and unveiling a 2โmillionโtoken context window.
Project Suncatcher shows how space-based AI compute can harness solar power and terabit optical links to train massive models beyond Earthโs energy limits.
Discover how Gemini 4 Argon cybersecurity equips teams with Argon Shield, realโtime threat hunting and onโpremise options to accelerate remediation.
AISIโs study shows a 29.2% success rate for GPT-6 Astra supply-chain attacks in simulated CI/CD pipelines, exposing serious risks even with scope limits.
Googleโs AI video co-director uses a multiโagent systemโCANVAS, AยฒRD, and VQQAโto keep longโform AIโgenerated videos consistent and free of flicker.
The Anthropic wet lab pairs Claudeโs AI with realโworld experiments, creating a closedโloop system that speeds earlyโstage drug discovery in the Bay Area.
A whiteโhat team leveraged the Claude Opus 5 hack to exploit a libheif CVEโ2026โ32882 flaw, breaching OpenAI and prompting a debate on AI model governance.
Trump dismisses AI safety warnings while leaders like Musk and Altman urge an AI development slowdown, underscoring the clash between competition and regulation.
ChatGPT Images 2.5 now generates sharper visuals up to 40% faster and introduces a Sketch tool for instant design tweaks, plus new templates and API options.
OpenAI and Dell Technologies announced a formal partnership to bring Codex into on-premises enterprise environments via the Dell AI Factory and Dell AI Data Platform. The path is real โ but it
Alibaba's Qwen3.6-27B is freely available for self-hosting, but its eight-GPU reference setup and a deteriorating open-source supply chain mean the barrier to production is higher than the free-access headline suggests. The model's
Alibaba's Qwen3.6-35B-A3B activates only ~3B of its 35 billion parameters at once, cutting inference costs without shrinking the architecture. Early assessments highlight strong agentic coding performance and a context window that stretches
Hugging Face has launched the Reachy Mini App Store with over 200 free, community-built applications for its $299 open-source desktop robot. No coding skills are required to build or use appsโbut creator
OpenAI has formally launched the OpenAI Deployment Company, a new enterprise AI unit backed by more than $4 billion in initial investment and the acquisition of AI consulting firm Tomoro. The venture
NVIDIA's Nemotron 3 Nano Omni multimodal model targets on-device inference at a moment when competitors are closing capability gaps and major customers are actively reducing their NVIDIA dependencies. The compact model addresses
DeepSeek released preview versions of its V4 model on Friday, offering two open-source variants that the company says rival leading closed-source systems from Google, OpenAI, and Anthropic. The release arrives with strong
OpenAI released GPT-5.5 on Thursday, claiming leads on 14 benchmarks and strong early results in coding and scientific research. The model's doubled pricing and tightly controlled Pro tier will determine how widely
Anthropic released Claude Opus 4.7, its most capable generally available model, with 13% coding gains, tripled image resolution, and improved agentic efficiency. The launch comes with an unusual caveat: the company's more
Meta launched Muse Spark on April 8, 2026 โ its first proprietary AI model built from scratch by the newly formed Meta Superintelligence Labs under 29-year-old Alexandr Wang. The model scores 52