SECTION · 34 stories
Releases & Models
Releases, new models, integrations — and what actually changes for operators.
Nvidia Launches the First CPU Built for AI Agents, and SpaceXAI Will Use It Even in Orbit
Nvidia announced on August 24, 2026 that SpaceXAI will deploy Vera, its first CPU designed specifically for artificial intelligence agents, to accelerate its next generation of agentic workloads, according to an official Nvidia announcement also distributed via GlobeNewswire and confirmed by StorageReview, WCCFTech, Seeking Alpha and StockTitan. SpaceXAI is also expanding Grok's AI infrastructure with the Nvidia Vera Rubin platform, scaling toward gigawatts of computing capacity, and plans to take the technology into space itself: the first generation Starmind AI satellite will be based on an NVIDIA Vera Rubin NVL72 system optimized for orbit. According to Nvidia, Vera completes tasks up to 1.8 times faster than x86 CPUs across agentic AI, reinforcement learning and data processing workloads.
Mistral Assembles a Coalition of European Companies to Fund Up to 1 Gigawatt of Sovereign Compute by 2030
Mistral AI announced three moves aimed at European digital sovereignty on August 11, 2026: regional inference endpoints letting customers choose whether their data is processed in Europe or the United States, a Priority Tier in public preview with a 99.5% uptime guarantee for mission critical workloads, and a coalition of European companies making multi year purchase commitments to fund up to 1 gigawatt of compute infrastructure in Europe by 2030, according to an official company announcement confirmed by VentureBeat, eWeek and other outlets. Companies including Amadeus, ASML, Capgemini, Caisse des Dépôts and CMA CGM have already signed on, underwriting an expected 200 megawatts of capacity by the end of 2027. Mistral is also starting to host third party open models on its platform, beginning with Z.ai's GLM-5.2.
Nvidia Pays $6 Billion to Poolside to Build an Open Weight AI Model Rivaling OpenAI and Anthropic
Nvidia has struck a $6 billion licensing deal with startup Poolside to use its model building technology, called Model Factory, while also investing another $1 billion in the company at a $12 billion pre-money valuation, according to a Bloomberg report published on August 20, 2026 and confirmed by Benzinga, PYMNTS, Yahoo Finance and Forbes. More than 100 Poolside employees, including engineers, are expected to join Nvidia to work on its open weight Nemotron AI project, while Poolside continues operating independently under its three co-founders. The company says the goal is to advance artificial general intelligence as an open technology rather than one controlled by a few companies, a move aimed squarely at rivals including OpenAI, Anthropic, DeepSeek and Kimi.
AWS Expands Access to OpenAI's GPT-5.6 Models With Cross-Region Inference and Cybersecurity Models
AWS began offering cross-region inference for OpenAI's GPT-5.6 Sol, Terra and Luna models within Amazon Bedrock on August 17, 2026, according to an official AWS announcement confirmed by TechTarget. The feature automatically routes requests across multiple AWS regions to gain more processing capacity and reduce inference cost, with support for the Responses, Converse and Chat Completions APIs. The announcement comes a week after AWS made two OpenAI cybersecurity models available on Bedrock, named Daybreak Red and Daybreak Blue, and weeks after cutting the price of the lightest model in the GPT-5.6 family by up to 80 percent.
All stories
34OpenAI Launches ChatGPT for Teens With Default Protections and Parental Controls
OpenAI began rolling out ChatGPT for Teens globally on August 18, 2026, a version of the product built for users ages 13 to 17 with content protections on by default, stronger restrictions on conversations about self-harm, suicide, and romantic or sexual topics, and a new study mode called Study Mode. According to the company's official blog and reporting from TechCrunch and Technology.org, parents can link accounts, set quiet hours and decide whether Study Mode is on by default, but cannot access the content of their children's conversations except in rare situations involving serious safety risk.
Cerebras Launches CS-4, a System Promising Up to 30 Times Faster AI Inference Than GPUs
Cerebras Systems announced the CS-4 on August 19, 2026, its first rack-scale system combining three WSE-3 Turbo wafers running in parallel at 2.8 GHz, double the clock speed of the original WSE-3. According to the company's official announcement and coverage from The Next Platform and ServeTheHome, the CS-4 delivers up to 30 times more tokens per second per user than GPU-based solutions and up to 10 times more throughput per watt than the CS-3, its predecessor, while using 50 percent fewer rack components. First units are expected to ship in the third quarter of 2026.
xAI Brings Grok 4.6 to GitHub Copilot Across Eight Development Surfaces
Two days after launching Grok 4.6 on August 12, 2026, xAI announced on August 14 that the model had arrived in GitHub Copilot, selectable from the model picker across eight surfaces: VS Code, Visual Studio, the Copilot CLI, Copilot's cloud agent, the Copilot app, JetBrains IDEs, Xcode and Eclipse. Built for agentic coding and multi step tasks, the model reaches Copilot's Pro, Pro+, Max, Business and Enterprise plans, though on the corporate tiers it ships off by default and requires an administrator to turn it on.
Google Launches Gemini 3.7 Flash, a Cheap Model for Code and Agents
Google launched Gemini 3.7 Flash on August 13, 2026, a workhorse model built for software engineering, AI agents and multi-step task execution. According to Google's official blog and SiliconANGLE, the model scored 65.3% on the DeepSWE v1.1 benchmark, up from 49.0% for Gemini 3.6 Flash. Introductory pricing of $0.75 per million input tokens and $3.75 per million output tokens runs through December 31, 2026, alongside a context window of roughly 1.05 million tokens. Bloomberg reports the launch comes as Google's next-generation flagship model remains delayed.
Google Launches Pixel 11 With Tensor G6 Chip and Gemini Acting Across 40+ Apps
Google unveiled the Pixel 11 lineup on August 12, 2026, featuring the 2 nanometer Tensor G6 chip and Gemini Intelligence carrying out multistep tasks across more than 40 apps, such as autonomously ordering food or booking a ride. According to Google's official blog, TechTimes and 9to5Google, prices start at $899, with the increase attributed to higher chip costs, and retail availability begins August 20.
DeepSeek Moves V4 Pro Out of Preview, Targets AI Agents
DeepSeek launched the production version of V4 Pro, dubbed 0813, on August 12, 2026, ending nearly four months in preview. According to Unite.AI and Tech Times, the Chinese model, with 1.6 trillion total parameters and 49 billion active per token, focuses on agent capabilities, with significant jumps on benchmarks such as DeepSWE and Terminal Bench, keeping preview era pricing for now.
Microsoft Preps Maia 300 AI Chip Reveal for the Fall, Aiming to Cut Nvidia Dependence
Microsoft plans to publicly unveil the Maia 300, its new AI accelerator chip, as soon as this fall, possibly in September 2026, according to reporting from Benzinga, TechStartups and AI Weekly published on August 10. The company has already discussed with Taiwanese manufacturer TSMC reserving capacity for more than 300,000 units of the chip for delivery in 2027, with ambitions of reaching more than 1 million units in total, a move that signals Microsoft's push to reduce its dependence on Nvidia GPUs.
Zuckerberg Publishes Superintelligence Manifesto as Meta Launches Open Source Muse Glimmer
On August 10, 2026, Mark Zuckerberg published the essay 'The Future is for Everyone,' roughly 6,500 words long, arguing that superintelligence should not be concentrated among a few labs, governments or companies. The same day, Meta released Muse Glimmer, a lightweight model under a permissive open license, and expanded access to Muse Spark 1.2. The story was reported by CBS News, Axios, ABC News/AP, Las Vegas Sun, PYMNTS, AOL and The Washington Post.
xAI Launches Grok Imagine Image 2.0 With Region Editing and Five Reference Images
xAI launched Grok Imagine Image 2.0 on August 7, 2026 as Quality Mode on grok.com/imagine and inside its iOS and Android apps. The update brings region based editing with a magic wand style tool, background removal with transparent export, and composition from up to five reference images in a single generation. According to xAI itself and independent coverage from Unite.AI and TestingCatalog, the model ranks second on the Arena leaderboards for image generation and editing, behind OpenAI's gpt-image-2.
Meta Launches Muse Code, Its First AI Coding Agent
Meta launched Muse Code, its first AI coding agent, in public beta on August 5, 2026, for macOS and Linux. Built on the new Muse Spark 1.2 model, the terminal based tool plans, writes, and reviews whole engineering tasks, with pay as you go pricing and a cheaper contributor tier in exchange for training data. The launch puts Meta in direct competition with Anthropic's Claude Code and OpenAI's Codex.
Grok 4.6: xAI Launches New Model, Already Preps Grok 4.7
xAI released Grok 4.6 on August 7, 2026, hitting the deadline Elon Musk had set. The new model keeps Grok 4.5's 1.5 trillion parameter foundation but adds major gains through upgraded supervised fine tuning and reinforcement learning. xAI is already signaling Grok 4.7, built on a larger 2.1 trillion parameter base, for the coming weeks.
OpenAI Opens Free ChatGPT Access to 100,000 Academic Researchers Through 2027
OpenAI announced on July 29, 2026 a program that will give 100,000 academic researchers free access to its most advanced models through 2027, starting with 10,000 researchers this summer. The package includes a dedicated ChatGPT workspace, usage equivalent to the $200 a month Pro plan, and is part of a commitment exceeding $250 million in external scientific research.
Microsoft Launches MAI-Cyber-1-Flash, an AI Model That Beats Rivals on Cybersecurity at Half the Price
Microsoft launched MAI-Cyber-1-Flash on July 27, 2026, its first cybersecurity specialized AI model, alongside the Project Perception platform. Running inside MDASH, the model scored 95.95% on the CyberGym benchmark, beating Anthropic, Google and OpenAI, at half the cost of market leaders.
Kimi K3 Becomes the World's Largest Open Weight AI Model as Moonshot Releases It for Free
Moonshot AI released the open weights for Kimi K3 on July 27, 2026, a 2.8 trillion parameter model that becomes the largest open weight model ever published. With a Modified MIT license and immediate hosted access through Together AI and Modal, the launch lands amid political tension in the United States over Chinese origin open models.
Mistral Confirms a New "Fat but Sparse" Open Model Family, With Early Access Starting in July
Mistral AI CEO Arthur Mensch confirmed a new open weight model family using a mixture-of-experts architecture he describes as "fat but sparse," with early access opening in July 2026 for research, government and industry partners. The company has not yet disclosed parameter count, benchmark results or licensing terms, but Mistral's annual recurring revenue has already topped $400 million.
OpenAI Opens ChatGPT Health to All US Adults One Day After a Lawsuit Sought to Block It
OpenAI opened ChatGPT Health to every adult in the United States on July 23, 2026, one day after a former Florida pastor asked a California court to block the feature. More than 300 million people already ask ChatGPT health questions every week, and the feature now draws on connected medical records and Apple Health data during conversations, with permission. It is the second lawsuit in three months accusing the chatbot of giving dangerous medical advice.
Anthropic Launches Claude Opus 5 at Same Price as Previous Generation, Nearing Fable 5 Performance
Anthropic launched Claude Opus 5 on July 24, 2026, a more cost efficient model for everyday tasks that keeps the same pricing as Claude Opus 4.8 while approaching, and on several benchmarks surpassing, Claude Fable 5. It is the fourth Claude 5 generation release in under two months, confirming a shift toward frequent incremental updates rather than rare blockbuster launches.
Google Ships a Trio of Gemini Flash Models While 3.5 Pro Stays MIA and Gemini 4 Training Begins
On July 21, 2026, Google launched Gemini 3.6 Flash, Gemini 3.5 Flash-Lite and Gemini 3.5 Flash Cyber, while confirming Gemini 3.5 Pro still has no broad launch date and revealing it has begun training Gemini 4, according to TechCrunch and Google.
Musk Announces 2 Trillion Parameter Grok 4.6 to Answer Kimi K3
On July 18, 2026, Elon Musk revealed that xAI is training Grok 4.6, a 2 trillion parameter model expected to finish its first training cycle this week. The announcement comes two days after Kimi K3, from China's Moonshot AI, took the top spot on a coding leaderboard and put pressure on Western models over cost.
Alibaba Previews Qwen3.8-Max, Says the Model Trails Only Claude Fable 5
On July 19, 2026, during WAIC in Shanghai, Alibaba unveiled Qwen3.8-Max-Preview, a 2.4 trillion parameter multimodal model the company positions as the world's second strongest, trailing only Anthropic's Claude Fable 5. Access is already open via Token Plan at 10% of standard pricing, but still without a published benchmark table, model card or license.
ZTE Launches NaviX Ultra, the World's First Smartphone With an AI Agent Built Into the OS
ZTE unveiled the NaviX Ultra at WAIC in Shanghai, a Nubia sub-brand phone running ByteDance's Doubao assistant that promises to complete tasks across multiple apps on its own. The first batch of 30,000 units sold out fast and already doubled in price on the resale market.
Google Delays Gemini 3.5 Pro Launch by Months After Retraining Falls Short
Google has once again pushed back the broad rollout of Gemini 3.5 Pro, promised for June 2026 by Alphabet CEO Sundar Pichai at Google I/O. According to Bloomberg, cited by 9to5Google on July 16, 2026, a late June retraining effort failed to fix the model's shortcomings in coding and complex reasoning.
Mira Murati Launches Inkling, Thinking Machines' First Open Model, and Admits It's Not the Strongest on the Market
Thinking Machines, Mira Murati's startup valued at $12 billion, launched Inkling on July 15, 2026, a 975 billion parameter open model the company itself says isn't the market's strongest.
Moonshot AI's Kimi K3 Arrives as the World's Most Advanced Open Model, at a Fraction of the Price
Moonshot AI launched Kimi K3 on July 16, 2026, a 2.8 trillion parameter model that ranked fourth on the Artificial Analysis Intelligence Index, priced well below Western competitors, with open weights expected by July 27.
SpaceXAI announces Grok 4.5, an "Opus-class" model cheaper than rivals
On July 8, 2026, SpaceXAI (formerly xAI) announced Grok 4.5, a model the company describes as comparable in class to Claude Opus 4.7, with public availability expected on July 9. The model costs $2 per million input tokens and $6 per million output tokens, undercutting rival pricing.
Anthropic launches Claude Sonnet 5, its most agentic Sonnet model yet
Anthropic launched Claude Sonnet 5 on June 30, 2026, now the default model on the Free and Pro plans, with a 1 million token context window and promotional pricing through August.
LangChain launches Deep Agents and updates LangGraph to version 1.2.4
LangChain launched Deep Agents in March, a higher-level framework built on LangGraph, while LangGraph itself reached version 1.2.4 on June 2.
OpenAI previews GPT-5.6 with restricted access
OpenAI announced a limited preview of the GPT-5.6 family (Sol, Terra and Luna) on June 26, currently available to only about 20 companies approved by the US government.
xAI launches Grok V9-Medium, a coding-focused model
Grok V9-Medium, a 1.5-trillion-parameter model trained on real Cursor session data, arrived on Grok and SuperGrok on June 16.
Anthropic launches Claude Fable 5 and Mythos 5
Anthropic launched two frontier models on June 9: Fable 5, the public version with a reinforced safety layer, and Mythos 5, which shares the same underlying weights.