Anthropic has released its latest frontier AI model, Claude Sonnet 4.5, built for high-performance coding and building complex agents. Anthropic claims it is “the best coding model in the world” and “the best model at using computers,” with substantial gains in reasoning and math. Claude Sonnet 4.5 is distinctive in that it can build “production-ready” applications and is not limited to prototypes, Anthropic says, describing it as a major gain over previous models. The new model also excels in research as well as enterprise basics like cybersecurity and finance, per Anthropic. Continue reading Anthropic Claude Sonnet 4.5 Good at Coding and Collegiality

OpenAI is taking ChatGPT beyond the question-and-answer phase with the introduction of ChatGPT Pulse, which has agentic abilities, proactively doing research to deliver personalized updates based on chats, feedback, and connected apps like calendar. Users can curate the results by indicating what’s useful. ChatGPT will be working while you’re not, delivering up to 10 “briefs” designed to help start your day. Available in preview to ChatGPT Pro users on mobile, OpenAI plans to “learn and improve from early use before rolling it out to Plus, with the goal of making it available to everyone.” Continue reading OpenAI’s Agentic ChatGPT Pulse Aims to Help Start Each Day

OpenAI and Anthropic — rivals in the AI space who guard their proprietary systems — joined forces for a misalignment evaluation, safety testing each other’s models to identify when and how they fall short of human values. Among the findings: reasoning models including Anthropic’s Claude Opus 4 and Sonnet 4, and OpenAI’s o3 and o4-mini resist jailbreaks, while conversational models like GPT-4.1 were susceptible to prompts or techniques intended to bypass safety protocols. Although the test results were unveiled as users complain chatbots have become overly sycophantic, the tests were “primarily interested in understanding model propensities for harmful action,” per OpenAI. Continue reading Anthropic and OpenAI Report Findings of Joint AI Safety Tests

Netflix has issued partner guidelines outlining use of generative AI in content workflows. The guidelines emphasize circumstances in which the use of generative AI requires advanced written approval, such as altering a character or performance, any inclusion of generative AI in the final product or using AI in a way that materially impacts union work. Fabricated content that could be “mistaken for real events” is to be avoided. Use of AI for “ideation” is deemed generally acceptable if done within guidelines. Allowing models to train on input or output of material destined for Netflix is off-limits. Continue reading Netflix Publishes Partner Guidelines for GenAI Production Use

This week, DeepSeek-V3.1 dropped on Hugging Face. Media outlets immediately began citing benchmark scores that rival proprietary systems from OpenAI and Anthropic for a system that is available via a permissive license, facilitating wide access. The 685-billion parameter Mixture-of-Experts (MoE) model has 37 billion active parameters and is designed for efficiency. It builds on DeepSeek-pioneered processes like multi-head latent attention (MLA) and multi-token prediction (MTP) to optimize inference, enabling high-performance computing on both enterprise servers loaded with H100 GPUs and consumer hardware like a Mac Studio or comparably powered PC. Continue reading DeepSeek-V3.1 Offered with Improvements in Speed, Context

Hisense Channels is a new FAST service launched by TV manufacturer Hisense and Xumo, the streaming joint venture between Charter Communications and Comcast. Boasting in excess of 200 premium channels from content providers including Scripps, Stingray and Radial Entertainment, Hisense Channels is debuting on Hisense smart TVs, where viewers can access it from the TV’s homepage screen — no need for a subscription or download. The move is an expansion of the existing partnership between Xumo and Hisense, which is interested in simplifying content discovery with built-in streaming access. Continue reading Xumo Powers Hisense Channels’ 200 FAST Content Offerings

Google has introduced a new ultra-light model called Gemma 3 270M ideal for smartphones and other on-device use cases. The open-source model is power-efficient and small enough to run locally in the absence of an Internet connection, as Google demonstrated in internal tests using a Pixel 9 Pro SoC. With just 270 million parameters, Gemma 3 270M is a fraction of the size of flagship LLMs, which typically have billions of parameters. While Google’s new model was not made for complex conversational use, it is “designed from the ground up for task-specific fine-tuning with strong instruction-following.” Continue reading Google Says New Gemma 3 Is Ideal for Mobile, Edge Devices

AI audio firm ElevenLabs has launched Eleven Music, which lets businesses and individuals generate studio-caliber music using natural language prompts. Users can generate tracks in any genre or style, even adding vocals in different languages. The Eleven Music model was developed in partnership with music licensing firm Merlin and independent publisher Kobalt. Artists and songwriters from the two groups will participate in the development of Eleven Music Pro, a subsequent model planned for release in the coming months. The company says it built-in guardrails to protect rightsholders. Continue reading ElevenLabs Debuts Eleven Music with Kobalt, Merlin Backing

Nvidia has unveiled the Blackwell Server Edition GPU designed for enterprise servers. The reveal was made at the ACM SIGGRAPH 2025 computer graphics conference, which started Sunday and runs through Thursday in Vancouver. The company also introduced a host of resources for robotics developers that include a new AI family called the Cosmos World Foundation Models, or Cosmos WFMs, which generate “physics-aware” videos. Notable among them is Cosmos Reason, an open and customizable 7-billion-parameter reasoning vision language model (VLM) for physical AI and robotics. Continue reading SIGGRAPH: Nvidia Touts Server Chip, Cosmos World Models

OpenAI is rolling a new foundation model, GPT-5, via API for developers and enterprise users in three branded sizes — gpt-5, gpt-5-mini and gpt-5-nano — “to give developers more flexibility to trade off performance, cost, and latency.” The company said Thursday that it is also making GPT‑5 available to all ChatGPT Plus, Pro, Team and Free tier users. Enterprise and Education tier users are promised access this week. While GPT‑5 in the API platform is the reasoning model that powers maximum performance in ChatGPT, “GPT‑5 in ChatGPT is a system of reasoning, non-reasoning, and router models,” OpenAI explains. Continue reading OpenAI Announces Launch of GPT-5 Model Across All Tiers

OpenAI has reportedly secured another $8.3 billion in funding, for a valuation of $300 billion. The new investment is a major coup for the firm, coming months ahead of schedule in its plan to secure $40 billion in funding by the end of the year. The company previously secured a $30 billion commitment from SoftBank if certain goals are achieved by the start of 2026. The funding news comes as ChatGPT is close to reaching 700 million weekly active users, including a significant jump in new users over the last several months. The company has also experienced an increase in paying subscribers as more enterprises and educators turn to AI. Continue reading OpenAI Raises $8.3B as ChatGPT Nears 700M Weekly Users

Google has upgraded Gemini 2.5 with a feature called Deep Think that “could be a powerful tool in creative problem solving,” according to the company. Deep Think uses parallel thinking techniques, letting Gemini “generate many ideas at once and consider them simultaneously, even revising or combining different ideas over time, before arriving at the best answer.” Its ability to reason through highly complex problems makes Deep Think mode a powerful tool for researchers, Google says, adding that it also excels at coding. Deep Think is available in the Gemini app to subscribers of Google AI Ultra, priced at $250 per month. Continue reading Google Intros Deep Think Reasoning Model for AI Ultra Subs

New York-based Runway AI has introduced a sophisticated video model called Aleph that can perform a wide range of edits from text prompts — adding, removing and transforming objects; generating various angles on a scene; or modifying style and lighting, among other things. Aimed at streamlining post-production, Runway calls Aleph an “in-context video model,” meaning it is designed to work with existing visual material rather than generating imagery from scratch. Using Aleph puts storytellers just a prompt or two from turning that wide shot into an extreme close-up, or adding a new “next shot,” providing what Runway calls “endless coverage.” Continue reading Runway Aleph Provides Video Editors with ‘Endless Coverage’

Moonvalley, the AI startup behind Marey, a high-quality video generator trained exclusively on licensed content, has just put the product in general release. The credits-based subscription pricing ranges from $15 to $150 per month. In addition to ethical training on 1080p native video, Marey also takes a non-traditional approach on its user interface, eschewing prompts for what it says is a more creatively intuitive process. “Directors need precise control over every creative decision, plus legal confidence for commercial use. Today we’re delivering both,” says Moonvalley CEO and co-founder Naeem Talukdar. Continue reading Moonvalley’s Production-Tailored AI Marey Publicly Released

New Jersey-based cloud-computing startup CoreWeave has reached an agreement to acquire crypto miner Core Scientific in a vertical integration move that will see the AI infrastructure provider gain access to more than 1 gigawatt of U.S. data center capacity with an incremental 1 gigawatt of power available for expansion. The all-stock transaction, valued at $9 billion, is expected to close in Q4 pending regulatory approval. According to CoreWeave, the purchase will eliminate some $10 billion in upcoming lease fees, saving around $500 million annually starting in 2027 and helping to “future-proof” the company. Continue reading AI Provider CoreWeave to Acquire Core Scientific for $9 Billion