AI News Archive: June 16, 2026 — Part 4
Sourced from 500+ daily AI sources, scored by relevance.
- G7 leaders discuss 'trusted partners' access for cutting-edge US AI models, sources say
G7 leaders discuss 'trusted partners' access for cutting-edge US AI models, sources say The Straits Times
- Microsoft’s newest AI agent wants to take entire projects off your plate
Microsoft is bringing Copilot Cowork to businesses worldwide, promising an AI agent that can complete complex, multi-step tasks while offering flexible pay-as-you-go pricing.
Score: 63🌐 MovesJun 16, 2026https://www.digitaltrends.com/cool-tech/microsofts-newest-ai-agent-wants-to-take-entire-projects-off-your-plate/ - AI helping build better AI: How agents accelerate model experimentation
How AI agents accelerate model experimentation
- Databricks' Ali Ghodsi: 'Your processes have to change'
Companies that try to add AI into their existing workflows without a firm grip on costs and their data are in for trouble, especially when the model at the heart of your agents can disappear overnight.
Score: 63🌐 MovesJun 16, 2026https://www.thestack.technology/databricks-ali-ghodsi-your-processes-have-to-change/ - Meta AI mode 📱, Factory 2.0 👨💻, Sakana’s autonomous researcher 🐟
Meta AI mode 📱, Factory 2.0 👨💻, Sakana’s autonomous researcher 🐟
- Trump's AI export strategy runs into Trump's export controls
The Trump administration has made exporting American AI a key part of its plans for global AI dominance, but ad hoc policy decisions around the most advanced AI are threatening that effort. Why it matters: A flagship U.S. program designed to boost AI exports could be undermined by the very administration that created it. "The government's willingness to arbitrarily and abruptly remove America's best models from all foreign use shows that the strategy behind the AI Export Program is no longer relevant to decision makers in the U.S. government," Dean Ball, a former AI adviser in the Trump administration, told Axios. Driving the news: The Trump administration slapped export controls on Anthropic's newest model Fable 5 due to disagreements over whether it is safe for deployment, causing Anthropic to pull access to it entirely. Administration officials and Anthropic staff continue to hash out their disagreements this week, with no solution yet. The big picture: The American AI exports program is a relatively new initiative that President Trump created in a July 2025 executive order . It's meant to bundle the infrastructure, tools and models into ready-to-deploy AI systems for allies and partners, and has been touted as a key part of the White House's AI policy goals, as Axios has previously reported. Those who are selected for the program will get expedited export control license reviews, prioritized access to U.S. federal credit programs, government-to-government advocacy abroad and dedicated interagency coordination. What they're saying: A tech industry source told Axios that there are "downstream consequences" to using export controls as a means of enforcing tech policy, setting new precedents for future oversight and licensing of tech releases. "Fueling perceptions that the US government could disable overseas access to an AI model or system only makes it more difficult to promote American AI exports," the source said. "Global customers will have a harder time committing to purchasing US-made AI." Other tech industry sources told Axios the Anthropic export control issue creates uncertainty by complicating relationships with allies at a time when there is a major focus on exporting U.S. technology abroad. "Given the interconnected nature of the AI tech stack, restrictions aimed at one layer or at one company in the stack can create unintended impacts for other parts of the stack," one of the sources said. "It definitely has a flavor of picking winners and losers, and the hope is that the U.S. government, in its efforts to promote American companies abroad, is going to do that consistently across the board, rather than picking up individual companies to prioritize over others," said Paul Lekas, vice president of public policy at the Software and Information Industry Association. Yes, but: Other AI companies looking to participate in the American AI exports program may avoid the problems that have befallen Anthropic. The upside of joining the program hasn't gone away, said Joseph Hoefer, AI principal at Monument Advocacy. But companies will have to build contingencies in case a layer of their "tech stack" suddenly becomes unavailable due to a decision by the administration. "This could turn out to be a one-off: a specific action, in a specific situation, that the administration resolves and doesn't repeat," he said. The other side: The White House defended the move as part of an effort to "balance" AI innovation and national security, per spokesperson Kush Desai. The Commerce Department's International Trade Administration did not respond to requests for comment. What's next: Applications for the American AI exports program are due June 30, and how the White House handles its dispute with Anthropic could shape whether companies feel confident participating.
Score: 63🌐 MovesJun 16, 2026https://www.axios.com/2026/06/16/trump-ai-export-strategy-export-control - Prompt Hungary: The impact of AI on the competitiveness of the economy
AI could become a powerful growth engine for the country by helping overcome the economy’s challenges and limitations, and generating improved productivity and lasting competitive advantage.
- SearchLeak attack turns Copilot Enterprise into an insider threat
SearchLeak attack turns Copilot Enterprise into an insider threat Computing UK
Score: 62🌐 MovesJun 16, 2026https://www.computing.co.uk/news/2026/security/searchleak-attack-turns-copilot-enterprise-into-an-insider-threat - Lotte pushes AI agents across workplace
Lotte Group Chairman Shin Dong-bin took part in the group's "CEO AI Academy" earlier this month, building AI-powered services and developing an AI agent himself as the conglomerate accelerates its artificial intelligence transformation. The two-day program, held June 5-6, brought together the CEOs of 50 Lotte affiliates to review the group's AI transformation strategy and discuss future business applications of AI. During the training, Shin created AI services using vibe coding and developed an
- Databricks says it solved the decades-old data pipeline problem that's been slowing AI agents
For decades, data professionals have struggled with the challenge of managing both operational and analytical databases in a unified approach that doesn't introduce latency and performance degradation. Agents made the problem structural. A system that reasons continuously and acts on live data cannot tolerate a pipeline between itself and the information it needs to act on. At the Data + AI Summit on Tuesday, Databricks announced two products aimed at collapsing that infrastructure. Lakehouse//RT delivers millisecond query latency directly on governed Delta and Iceberg tables, eliminating the dedicated real-time serving tier that enterprises have maintained alongside their lakehouses. LTAP, short for Lake Transactional/Analytical Processing, stores Postgres-native transactional data in Delta and Iceberg format from the point of write, removing the ETL pipelines that have connected operational and analytical systems for decades. Reynold Xin, co-founder of Databricks, described a simpler data stack as "the holy grail for agents" in a briefing with VentureBeat, arguing that as users vibe code more applications, the agents reasoning analytically on top of those apps need the underlying infrastructure out of the way to move fast. "The agents really prefer a much simpler stack, because they can move way faster," he said. LTAP bets on storage-layer unification where HTAP tried engine convergence Many vendors have tried various approaches over the decades to unify analytical and transactional data. Back in 2014, analyst firm Gartner coined the term HTAP, an acronym that stands for Hybrid Transactional/Analytical Processing as a way to describe vendors that attempted to unify the two types of databases. Vendors including MemSQL (now known as SingleStore ) SAP HANA and Oracle's MySQL Heatwave are among many HTAP vendors in the market. LTAP is Databricks' answer to HTAP, using the Lakebase architecture to unify data at the storage layer rather than the engine level. Lakebase is Databricks' serverless cloud-based PostgreSQL database service that became generally available in February. "HTAP to us is kind of more of a failure of the industry rather than a success," Xin said. The LTAP approach goes to the storage layer instead of the query layer. Lakebase previously stored Postgres data in Postgres format on object storage, requiring conversion before the Lakehouse's analytical engines could use it efficiently. With LTAP, transactional data lands directly in Delta or Iceberg format, sharing the same copy that analytical workloads read. Postgres remains the transactional engine. Spark and the Lakehouse remain the analytical engine. "The whole point is, hey, you use the best tool for the job at the query engine level, we just make sure underlying storage is a single copy of the data," Xin said. The central engineering challenge is latency. Object storage carries response times in the seconds range, far too slow for OLTP workloads that require sub-millisecond performance. Lakebase handles this through a caching layer between Postgres compute instances and object storage. The key design decision is where the column conversion happens: idle CPU capacity in that caching layer performs the row-to-column conversion before data lands in object storage. "When you convert data from row to column, it compresses more than 10 times, typically, so now you substantially reduce the network cost of that basic caching layer between that caching layer and the object stores," Xin said. Lakehouse//RT delivers millisecond query latency on live lakehouse data without a separate serving tier Lakehouse//RT is Databricks' answer to the dedicated real-time serving tier — the separate system enterprises have maintained alongside their lakehouses to handle low-latency queries, at the cost of data copies, split governance and pipeline complexity agents cannot work around. Key capabilities of Lakehouse//RT include: Reyden compute engine: Built specifically for high-concurrency, low-latency serving, Reyden queries Delta and Iceberg tables directly without moving data out of the lakehouse. Latency and throughput: Lakehouse//RT delivers sub-100ms latency at 12,000 queries per second, with response times as low as 10ms on smaller datasets and up to 16x better performance than existing dedicated serving stacks. Governance and data access: Every query runs within Unity Catalog's governance framework with no separate permissions layer, no data copies and no ingestion pipelines. Analysts see the agentic framing and open format approach as the real differentiators The problem both products address is well-documented among enterprise data teams, but analysts draw a distinction between the pain point and the specific claim Databricks is making. "Enterprises have had HTAP, streaming, cloud warehouses, and operational stores for years," Stephanie Walter, Practice Leader for AI Stack at HyperFRAME Research, told VentureBeat. "What is different is the agentic AI framing." Walter noted that agents need live operational data, historical context, governance, retrieval, and write-back in the same workflow. "That is a strong architecture argument, but Lakebase still has to prove it can meet the latency, reliability, and operational maturity CIOs expect," she said. Mike Leone, analyst at Moor Insights and Strategy, said the path to genuine differentiation is more specific than the unification concept itself. He also noted that open analytics on a data lake is table stakes now, with many vendors providing some sort of service. "The less common move is letting the transactional writes land in open formats too, so the operational database isn't sitting in a proprietary box while only the analytics half is open, "Leone told VentureBeat. He added that the open format approach, paired with Lakehouse//RT querying live data directly off the lake, is what gives the architecture a credible case for retiring a whole row of specialized systems. The technical claim that will face the most scrutiny is also the most central one. "The piece I'd still want their engineers to walk through is how both engines truly share one copy without a quiet conversion step doing the syncing in the middle," Leone said. What this means for enterprises For data engineers evaluating their stack for agentic workloads, the question is no longer which best-of-breed tool to run for each job — it's whether running separate tools at all is still defensible. Enterprises that built separate operational databases, real-time serving tiers and analytical lakehouses could previously treat the gaps between them as a maintenance burden. Agents surface those gaps as an operational risk: a system reasoning across governance boundaries will find the inconsistencies faster than any human team. The market is moving away from specialized serving layers faster than most vendor roadmaps anticipated. According to VB Pulse Q1 2026 , a three-wave longitudinal survey of 100-plus employee organizations, hybrid retrieval intent tripled from 10.3% to 33.3% across the quarter while standalone vector database adoption declined across every tracked vendor. The same consolidation logic is now hitting the real-time serving tier. The traditional approach — best-of-breed tools for each workload type, pipelines between them — was built for human-speed analytical consumption. Agent workloads don't tolerate that architecture. "The pain they're pointing at, all the copying and syncing between operational and analytical systems, is real and expensive, and anyone running this at scale feels it," Leone said.
- AI Is Ready. Enterprises Are Not. Vendors Need to Fix It.
We are projecting global IT spending on AI to reach $409 billion in 2026, roughly 53% year-over-year growth, and on track to reach $700 billion by 2029. That is not a trend. That is a structural transformation of the global technology economy, playing out in real time. And yet, for all that investment, the enterprise […] The post AI Is Ready. Enterprises Are Not. Vendors Need to Fix It. appeared first on IDC .
Score: 61🌐 MovesJun 16, 2026https://www.idc.com/resource-center/blog/ai-is-ready-enterprises-are-not-vendors-need-to-fix-it/ - Make no mistake: the U.S. now has a licensing regime for frontier AI
Make no mistake: the U.S. now has a licensing regime for frontier AI Fortune
- Tesla Cybercab Specs Are Public — But Questions Remain
Tesla Cybercab specs have been revealed via an EPA filing. It’s a 15-page document, but here are some of the key facts and figures: Battery Capacity: 326-volt system, 146 Ah — probably around 50 kWh energy storage capacity Electric Motor: 163 kW (219 hp), front-mounted AC permanent magnet motor Curb ... [continued] The post Tesla Cybercab Specs Are Public — But Questions Remain appeared first on CleanTechnica .
Score: 60🌐 MovesJun 16, 2026https://cleantechnica.com/2026/06/15/tesla-cybercab-specs-are-public-but-questions-remain/ - Databricks unveils CustomerLake, its agentic CDP
After weeks of speculation, Databricks entered the martech space with an agentic CDP built for an agentic era of marketing and shopping. The post Databricks unveils CustomerLake, its agentic CDP appeared first on MarTech .
- France to invest €655 million in AI, set up common chatbot for all state services
The French government will create a public health chatbot for state-owned health insurance Ameli agency. "We can either be subjected to this (Artificial intelligence) revolution, or we can lead it," he said in a post on X.
- Inside the cloud's new agentic AI-ready, Arm-powered foundation
PARTNER CONTENT: From hyperscalers to enterprises, performance-per-watt and system-level efficiency are redefining the cloud compute foundation
- AI’s impact on cognitive ability: MIT study reveals more troubling data
Yet another study shows that the more you let artificial intelligence do the thinking for you, the less capable you are on your own. This time, researchers at MIT tested how relying on AI to tell fake news apart from the truth impacted users’ ability to identify misinformation on their own. Treating AI chatbots as a news source is increasingly common, particularly among young people. Recent reports from Pew Research Center show that one in five teenagers in the U.S. get their news from chatbots, while one in five adults under age 50 report using AI for their news at least sometimes. The study out of the MIT Media Lab tracked 67 participants over four weeks as they evaluated news headlines and images, saying whether they believed they were real or fake, sometimes with the assistance of an AI chatbot. When they had the chatbot’s help, participants were 21% more accurate in finding fake news with the AI’s help—but at the end of the study, a troubling side effect emerged. By week four of the study, participants’ unassisted ability to identify fake news had declined by 15 percentage points compared to their scores before the experiment started. Their confidence, however, increased: a quarter of participants said they felt their detection abilities had improved, even as their performance got worse. Anku Rani, the co-lead author of a paper about the study, told MIT News that the results reflect people’s misplaced trust in AI. “Users get excited about these ‘magical’ LLMs, but forget that they’re just statistical models that predict the next ‘token’ in a sequence,” Rani said. “Many impressive behaviors emerge from scaling this, but it comes with real limitations, both in what the model can reliably generate and in its broader impact on the people using it.” AI and cognitive decline This study is far from the first to show that relying on AI negatively impacts cognitive ability. A recent May study that saw using AI for just a 10-minute period left study participants less able to solve math problems and SAT-style reading questions . Then there are studies about doctors who lost their ability to detect cancer independently , data workers whose critical thinking skills deteriorated , and essay writers whose brain activity declined , all after becoming reliant on AI assistance to complete tasks. Taken together, these studies point to what’s known as the “AI dependency paradox,” where humans’ skills initially improve when assisted by AI, only to fall below their previous baseline when that AI help is removed. A smarter way to use AI Though the study should give pause to those relying on AI to distinguish real news from fake, it adds that there are still ways AI can help without sacrificing your own judgment skills in the process. Valdemar Danry, the study’s other co-lead author, suggested AI conversations based in the Socratic method—interactions where the AI asked guiding questions to steer participants toward the correct answer, rather than outright providing it—could help participants build the skills to identify fake news on their own, even when the AI was removed. “AIs that ‘tell’ by providing direct answers are more likely to foster reliance, while those that ‘ask’ via Socratic questioning are better at engaging someone to actually learn how to discern the truth on their own,” Danry said. “But it’s very much a trade-off between speed and effort.” “There’s a lot of work to do in making sure that we don’t just fully offload critical tasks that we want to be able to keep on doing to these models,” he added. “We need to develop a new kind of AI literacy.”
- Databricks strikes deal to buy Panther Labs
In cyber security push.
- AIRS Medical Welcomes Strategic Growth Investment from TA Associates to Accelerate Global Growth in AI-Powered MRI Solutions
Investment supports global expansion and continued innovation in AI-powered MRI technologies that help healthcare providers improve imaging efficiency, increase capacity and expand patient access
- Dubai startups launch 32 new apps and 60% are powered by AI
Dubai startups launch 32 new apps and 60% are powered by AI Arabian Business
Score: 60🌐 MovesJun 16, 2026https://www.arabianbusiness.com/business/technology/dubai-startups-apps-ai - IT 2030: The AI Future of Work and Tech
IT 2030: The AI Future of Work and Tech Gartner
- ABB Robotics Enters Partnership to Advance Physical AI
The collaboration with Psyonic will focus on improving robot training systems to support physical AI rollout.
Score: 60🌐 MovesJun 16, 2026https://aibusiness.com/robotics/abb-robotics-enters-partnership-advance-physical-ai - Humanoid Robot Prices Crash Below 10,000 RMB as Mass Adoption Begins
Humanoid robot prices have plunged below 10,000 RMB for the first time, driven by China's mature supply chain and a strategic shift toward data collection over industrial-grade performance.
Score: 60🌐 MovesJun 16, 2026https://pandaily.com/humanoid-robot-prices-crash-below-10000-rmb-jun2026 - A.I. Boom Ignites Asian Chip Companies
They make much of the gear that goes into giant data centers. Demand for their products is shifting the balance of tech power.
Score: 60🌐 MovesJun 16, 2026https://www.nytimes.com/2026/06/16/technology/taiwan-south-korea-ai-chips.html - Qualcomm reveals flagship XR processor and new framework for AI glasses
Qualcomm's new Snapdragon Reality Elite platform brings faster AI processing, sharper visuals, longer battery life, and improved tracking to the next generation of XR headsets and smart glasses.
Score: 60🌐 MovesJun 16, 2026https://www.digitaltrends.com/computing/qualcomm-reveals-flagship-xr-processor-and-new-framework-for-ai-glasses/ - Meta uses your public Facebook posts for AI search
Meta's new Facebook AI Mode uses Muse Spark to generate answers — and will soon use public posts across Instagram, Facebook, and Threads.
Score: 60🌐 MovesJun 16, 2026https://mashable.com/tech/meta-muse-spark-ai-scrapes-public-posts-for-answers - Huawei Cloud, Thndr partner on cloud, AI technologies for Egypt’s financial services sector
For Thndr, the collaboration is expected to support ongoing efforts to enhance its digital investment platform through cloud and AI technologies
- Startup Backed by Ex-Google CEO Debuts Robot, LG Partnership
A startup backed by former Google Chief Executive Officer Eric Schmidt unveiled an industrial robot powered by artificial intelligence, as investors pile into the emerging field of humanoids.
- Qualcomm unveils its Snapdragon Reality Elite chip for next-gen AR headsets
Qualcomm's new Snapdragon Reality Elite promises a boost for AR and mixed reality devices.
Score: 59🌐 MovesJun 16, 2026https://www.engadget.com/2194353/qualcomm-unveils-its-snapdragon-reality-elite-chip-for-next-gen-ar-headsets/ - SoftBank launches cybersecurity product based on OpenAI models
SoftBank launches cybersecurity product based on OpenAI models Reuters
- AI Agents Are Your New Customer. But Can You Target and Grow Their Trust in Your Brand?
AI agents don’t browse your site or respond to marketing narratives. They retrieve, validate, and surface structured information to answer engines and consumer agents. That creates a fundamental shift towards business-to-agent (B2A) marketing strategies: how do marketing leaders target agents and grow their trust to retain their loyalty? To gain competitive advantage with machines – a.k.a. machine advantage – leaders must: 1. Look beyond getting content ready for agent visibility: Shape the agent’s context with a […]
- A robotic hand ‘talks’ to deaf and blind people. Here’s how it works.
A robotic hand ‘talks’ to deaf and blind people. Here’s how it works. The Boston Globe
Score: 58🌐 MovesJun 16, 2026https://www.bostonglobe.com/2026/06/16/business/tatum-robot-asl-sign-language/ - Exclusive: Mindbeam touts dramatic performance improvements in CPU-based AI inference
Two-year-old startup Mindbeam AI Inc. today released an open-source artificial intelligence inference framework designed to make large language models run more efficiently on standard consumer processors, a move the company says could reduce reliance on expensive graphics processing units for some AI workloads. Litespark-Inference is a software library that enables ternary large language models to run […] The post Exclusive: Mindbeam touts dramatic performance improvements in CPU-based AI inference appeared first on SiliconANGLE .
- Z.ai’s open-weights GLM-5.2 beats GPT-5.5 on multiple long-horizon coding benchmarks for 1/6th the cost
Today, Chinese AI startup Z.ai (formerly Zhipu AI) announced the immediate release of GLM-5.2 , a 753-billion parameter open-weights large language model (LLM) engineered specifically to dominate "long-horizon" autonomous coding and engineering tasks. Available immediately on Hugging Face , the Z.ai API , and more than 20 third-party coding environments, the model boasts a highly stable 1-million-token context window alongside enterprise subscription tiers starting at just $12.60 per month. In excellent news for cost and security-conscious businesses, z.ai has released GLM-5.2's core weights under an unrestricted MIT open-source license , allowing enterprises to download the model freely from Hugging Face, customize or fine-tune it to their liking, and run it potentially locally or via virtual machines for only the cost of their compute and electricity. This is an increasingly appealing option for enterprises, as state-of-the-art American proprietary models face an uncertain and potentially interrupted regulatory future, following the Trump Administration's export control directive last week prohibiting foreign nationals from using Anthropic's new Claude Fable 5 model (which that company responded to by taking the models in question entirely offline for all users). For enterprise technical decision-makers, z.ai's GLM-5.2 provides a highly capable path to host frontier-level AI locally, entirely bypassing the geographic fencing and commercial limitations. IndexShare re-uses one indexer for every four sparse attention layers, reducing compute needs Under the hood, GLM-5.2 operates with 753 billion parameters and introduces a major architectural optimization called "IndexShare". In standard massive language models, recalculating attention mechanisms across long documents is computationally exorbitant. IndexShare solves this by reusing the identical indexer across every four sparse attention layers. At the maximum 1-million-token context length, this single innovation reduces per-token compute FLOPs by a massive 2.9 times. The model also features an upgraded Multi-Token Prediction (MTP) layer for speculative decoding, which boosts accepted token length by up to 20% during inference. Additionally, Z.ai has implemented flexible, selectable "Thinking Modes". Users can toggle the model's reasoning effort between "Max," designed to push the limits of logical problem-solving, or "High," which strikes a careful balance between high-end performance and latency-sensitive token efficiency. State-of-the-art benchmarks for an open model, and matching, even beating proprietary leaders on some categories On industry-standard third-party benchmark tests, GLM-5.2 performs above most open source flagship models, even DeepSeek v4 and scores near or above its closed-weights rivals, OpenAI's GPT-5.5 and Anthropic's Claude Opus 4.8. The model particularly shines in agentic tool use and long-horizon software engineering tasks: SWE-bench Pro: GLM-5.2 scored 62.1, decisively beating GPT-5.5 (58.6) and its own predecessor, GLM-5.1 (58.4). FrontierSWE (Dominance): Designed to test long-horizon task completion, GLM-5.2 hit 74.4%, surpassing GPT-5.5 (72.6%) and finishing in a near-tie with Claude Opus 4.8 (75.1%). MCP-Atlas: On this tool-usage evaluation, GLM-5.2 achieved a 77.0, outscoring GPT-5.5 (75.3) and performing just shy of Claude Opus 4.8 (77.8). Humanity's Last Exam (w/ Tools): When equipped with external tools, GLM-5.2 reached a score of 54.7, coming out ahead of GPT-5.5 (52.2) and tracking closely behind Claude Opus 4.8 (57.9). PostTrainBench & SWE-Marathon: In extended, multi-hour engineering workloads, GLM-5.2 consistently topped GPT-5.5, scoring 34.3% against GPT-5.5's 25.0% on PostTrainBench, and 13.0% against GPT-5.5's 12.0% on SWE-Marathon. While GLM-5.2 trails Claude Opus 4.8 and GPT-5.5 slightly on raw Terminal-Bench 2.1 scores (81.0 versus 85.0 and 84.0, respectively), it significantly outscores Google's Gemini 3.1 Pro (74.0). Beyond traditional coding metrics, GLM-5.2 took an impressive first place on the crowdsourced design task benchmark Design Arena , beating out even the aforementioned state-of-the-art Claude Fable 5 with an ELO score of 1360. Furthermore, the impact of Z.ai's new selectable "thinking modes" is clearly visible in the data: under the "Max" effort level, GLM-5.2 pushes to peak intelligence, but utilizes nearly 85k output tokens per task. Switching to the "High" effort setting sacrifices only a few points in performance while effectively halving the required token output, providing a crucial optimization lever for latency-sensitive applications. Available via Coding Plans and API To operationalize the model, Z.ai launched the GLM Coding Plan , aiming squarely at developer workflows rather than simple chat interfaces. The plan offers out-of-the-box support for third-party U.S. and global agentic coding harnesses and tools including Claude Code, OpenClaw, Cline, Kilo Code, Crush, and Factory, among others. The Coding Plan pricing tiers (when billed annually) are highly competitive: Lite: $12.60 per month ($151.20 per year starting in the 2nd year), geared toward lightweight iteration on small repositories. Pro: $50.40 per month for day-to-day development on mid-sized repositories, offering 5x the usage allowance of the Lite plan. Max: $112.00 per month for heavy workloads, offering 20x the Lite usage and dedicated resources during peak hours. For enterprise developers integrating the raw model into their own applications, Z.ai's API pricing undercuts its Western rivals significantly while matching the exact rates of the previous GLM-5.1 generation. GLM-5.2 API access is priced at $1.40 per million input tokens and $4.40 per million output tokens , making it a mid-priced model globally, but about VentureBeat Frontier AI Model API Pricing Snapshot Sorted by total cost (input + output) from least to most expensive. Pricing shown is standard pay-as-you-go pricing per 1 million tokens. Model Input Output Total Cost Source MiMo-V2.5 Flash $0.10 $0.30 $0.40 Xiaomi MiMo deepseek-v4-flash $0.14 $0.28 $0.42 DeepSeek deepseek-v4-pro $0.435 $0.87 $1.305 DeepSeek MiniMax-M3 $0.30 $1.20 $1.50 MiniMax Gemini 3.1 Flash-Lite $0.25 $1.50 $1.75 Google Qwen3.7-Plus $0.40 $1.60 $2.00 Alibaba Cloud MiMo-V2.5 $0.40 $2.00 $2.40 Xiaomi MiMo Grok 4.3 (low context) $1.25 $2.50 $3.75 xAI MiMo-V2.5 Pro (≤256K) $1.00 $3.00 $4.00 Xiaomi MiMo Kimi-K2.6 $0.95 $4.00 $4.95 Moonshot/Kimi GLM-5.2 $1.40 $4.40 $5.80 Z.ai Grok 4.3 (high context) $2.50 $5.00 $7.50 xAI MiMo-V2.5 Pro (>256K) $2.00 $6.00 $8.00 Xiaomi MiMo Qwen3.7-Max $2.50 $7.50 $10.00 Alibaba Cloud Gemini 3.5 Flash $1.50 $9.00 $10.50 Google Gemini 3.1 Pro Preview (≤200K) $2.00 $12.00 $14.00 Google GPT-5.4 $2.50 $15.00 $17.50 OpenAI Gemini 3.1 Pro Preview (>200K) $4.00 $18.00 $22.00 Google Claude Opus 4.8 $5.00 $25.00 $30.00 Anthropic GPT-5.5 $5.00 $30.00 $35.00 OpenAI Claude Fable 5 / Claude Mythos 5 $10.00 $50.00 $60.00 Anthropic To further optimize costs for long-context workloads, Z.ai offers a cached input rate of just $0.26 per million tokens, alongside a limited-time offer for free cached input storage. The stark contrast between open-weights innovators and proprietary Western labs has not gone unnoticed by the developer community. On X, prolific AI observer Lisan al Gaib (@scaling01) argued that "frontier labs are absolutely scamming you on API pricing". The post noted that while massive open models like the 744-billion-parameter GLM-5.2 charge $4.40 per million output tokens and DeepSeek-V4-Pro (1.6 trillion parameters) charges just $0.87, proprietary models demand heavy premiums: Anthropic's Sonnet 4.6 and Opus 4.8 charge $15.00 and $25.00 respectively, while OpenAI's GPT-5.5 costs $30.00 for output. Highlighting that open-model developers are operating profitably without relying on the newest "fancy Blackwell chips," the commentator suggested that leading proprietary labs are "probably at 90%+ margins at this point". The beauty of the unmodified MIT License for enterprise use The most disruptive aspect of the GLM-5.2 release is its licensing. Z.ai released the model's weights under an MIT open-source license, establishing it as a "Pure Open" system. The company’s technical documentation explicitly notes that this license guarantees "no regional limits" and allows "technical access without borders". For enterprise technology leaders, an MIT license means the software can be used, modified, and commercialized without paying royalties or adhering to restrictive "acceptable use" governance policies common to dual-use licenses. It allows engineering teams to host frontier-level AI on their own sovereign infrastructure, entirely eliminating vendor lock-in. Warm reception among AI developers and toolmakers The developer reaction to the release has been immediate and overwhelmingly positive. The team behind Kilo Code confirmed day-one integration, posting on X: "GLM-5.2 runs in Kilo Code on day one. The 1M context window and Max effort mode are both live. Point your config at it and go!". Open-source coding environment Cline IDE echoed this sentiment on X , noting the economic advantage: "GLM-5.2 is the first open-weights model to cross 80% on Terminal-Bench, and beats every other open model available. It also beats Gemini, making it a frontier-level model for a fraction of the cost. Open weights is back. This model is a game changer. Available in Cline now!". Similarly, rival open source coding desktop agent Eigent AI also tested the model's new capabilities on complex agentic workflows, noting on X: "threw a real long-horizon task: research 30 companies across 6 sectors of the AI infrastructure stack, structure it into JSON, then build an interactive HTML report... where 5.2 pulls ahead: -> plans...".
- ‘Ni hao, Ah Bao’: China’s Alipay gets AI-agent overhaul to challenge big tech rivals
Ant Group has unveiled the biggest overhaul of Alipay in two decades, transforming its ubiquitous mobile-payment app into a native AI platform with autonomous agents, as it strives to dominate China’s next-generation internet gateway. A trial version of the flagship app now features an interactive assistant named “Ah Bao”, marking a strategic pivot from a traditional digital wallet into an AI-driven ecosystem. Accessible via a right swipe on the homepage, the new interface allows users to access...
- Siriraj, NCI unveil device to enhance surgical outcomes
The National Cancer Institute (NCI) and the Faculty of Medicine, Siriraj Hospital have introduced an advanced vascular connection device to improve surgical outcomes for cancer patients.
Score: 57🌐 MovesJun 16, 2026https://www.bangkokpost.com/thailand/general/3272081/siriraj-nci-unveil-device-to-enhance-surgical-outcomes - Why Kioxia is going easy on capex despite AI memory boom
Why Kioxia is going easy on capex despite AI memory boom Nikkei Asia
Score: 57🌐 MovesJun 16, 2026https://asia.nikkei.com/business/tech/semiconductors/why-kioxia-is-going-easy-on-capex-despite-ai-memory-boom - EU Commission keeps contact with Anthropic over decision to disable models in EU
EU Commission keeps contact with Anthropic over decision to disable models in EU Reuters
- The AI Pragmatist: Why SAS’ Udo Sglavo thinks most companies are solving the wrong problem
Udo Sglavo, Vice President of Applied AI and Modelling at SAS, has spent 25 years watching organisations fall in love with technology and forget the question they were trying to answer. In an exclusive conversation, he argues that the difference between AI leaders and laggards has nothing to do with the sophistication of their models — and everything to do with how they think The post The AI Pragmatist: Why SAS’ Udo Sglavo thinks most companies are solving the wrong problem appeared first on Express Computer .
- HPE AI Factory With NVIDIA Expands for the Era of Agents
Enterprises are moving agentic AI from proof of concept to production — and the next generation of AI factories are built for the era of agents. At HPE Discover Las Vegas, running through Thursday, June 18, NVIDIA and HPE are expanding the HPE AI Factory with NVIDIA, including NVIDIA Vera CPU and NVIDIA Agent Toolkit […]
- France to ditch Palantir’s AI data tools in favour of domestic provider
Move to ChapsVision is to avoid ‘strategic dependencies’, says PM amid concern about reliance on US-controlled tools France’s domestic intelligence service is to ditch AI data tools from the US tech company Palantir in favour of a domestic provider in an effort to avoid “strategic dependency”, the prime minister, Sébastien Lecornu, has said. “We must use our own AI models; we cannot accept new strategic dependencies in the digital sphere,” Lecornu posted on social media. “We cannot rely on tools developed by foreign powers. France must have its own tools.” Continue reading...
Score: 56🌐 MovesJun 16, 2026https://www.theguardian.com/world/2026/jun/16/france-ai-data-tools-palantir-chapsvision - Can AI help neurologists identify infant seizures more accurately?
Can AI help neurologists identify infant seizures more accurately? EurekAlert!
- Stanford's DeLM cuts multi-agent task costs 50% — without a central orchestrator
One of the assumptions behind today’s AI frameworks is that agents require a “boss” at the center; this orchestrator runs the show, routes requests, and makes sure the whole system doesn’t descend into chaos. That assumption may be wrong, and the cost of carrying it could be measured in inference dollars and coordination latency. A new Stanford framework called a decentralized language model, or DeLM, is built on the premise that agents can coordinate directly, without routing every update through a central controller. DeLM's shared knowledge base serves as a “common communication substrate” so that agents can build upon one another’s verified progress without having to route every interaction through a main agent to “merge, filter, and rebroadcast,” Yuzhen Mao and Azalia Mirhoseini, co-developers of the framework, explain in a research paper . It’s a system that’s not only possible, but desirable in certain instances. “Agents can build on prior findings, avoid repeated failures, preserve constraints, and recover detailed evidence only when needed.” The challenges of traditional multi-agent systems In a typical centralized multi-agent system, a main agent breaks tasks into subtasks, assigns them out to multiple sub-agents in parallel, waits for responses, merges and summarizes intermediate progress, then launches a next wave of orders based on collected context. While this is a natural way to scale LLM reasoning, the Stanford researchers argue that it scales poorly. Every useful finding, partial finding, and failure must be reported back to the main agent, which then determines what information to merge and rebroadcast to the agents below it. “As the number of subtasks grows, this controller becomes a communication and integration bottleneck,” Mao and Mirhoseini write. Further, the main orchestrator may “dilute, omit, or distort” useful information, leading to lost progress. This bottleneck also occurs in long-context reasoning scenarios. Once it receives reports back from subagents, a main agent will typically group related concepts, data points, and other materials together in an unsupervised learning loop. It may then pre-assign these "evidence clusters" to sub-agents before knowing what surfaced material is actually relevant or whether it’s combined correctly. When a subagent receives this insufficient context, it will essentially get confused and return to the main agent, kicking off another retrieval or delegation round. “This back-and-forth makes coordination slower, more iterative, and increasingly constrained by a single overloaded main agent,” the researchers write. What DeLM addresses and how it works DeLM, by contrast, is built around parallel agents, a shared context, and a task queue. Shared context is essentially a curated store of “gists,” or information summaries that other agents might find useful. These include verified and evidence-based findings alongside partial findings and documented failures; they also point to detailed evidence that agents can pull from based on their specific task. A task queue is then a set of subsequent pending subtasks that agents can claim independently. “Agents write compact, verified updates into a shared context that later agents can read directly,” the researchers write. Useful findings, failures, and constraints accumulate as a “shared problem state,” rather than passing through a central controller. The pipeline looks like this: Initialization: Inputs are broken into different work units and added to a queue; Parallel execution: Agents work independently and in tandem, pulling tasks and reading shared context as they progress. Compression and verification: Results are compressed into reusable “gists” that are checked against supporting evidence. Only gists that are fully verified are shared with the group. Additional work (if needed): When the queue is emptied, the last agent to return an answer inspects all the shared context to determine whether further work is required. Final step: The last agent determines that no more steps are required and returns the final answer. Agents “exchange progress through shared state, asynchronously claim ready tasks, and scale more adaptively as the number of subtasks grows,” the researchers explain. How DeLM performs in the wild With DeLM, agents can avoid redundant exploration; reuse and build on each other’s discoveries and failures; and focus on unresolved issues. The framework can be particularly useful in software engineering test-time scaling, when models are given time to “think” to improve their reasoning and problem-solving capabilities. Different agents can explore their own hypotheses or pursue reasoning paths in parallel, while still sharing intermediate progress. One example is concurrent de-bugging. DeLM is also suitable for long-context reasoning and multi-document question-answering; agents can simultaneously examine their own evidence clusters (collections of papers, code, or other materials) at the same time, while maintaining a “global compact view” of accumulated evidence. The researchers contend that it makes agentic tasks more accurate and significantly cheaper. This is backed by its performance on real-world benchmarks: On SWE-bench Verified — which evaluates how well AI models and agents solve real-world software engineering problems — it performed 10.5% better than the strongest baseline and reduced cost per task by roughly 50%. But it can go beyond coding: On LongBench‑v2 Multi‑Doc QA — which assesses LLMs’ ability to handle long-context, real-world problems — DeLM had the highest accuracy across four model families, including GPT‑5.4, Claude Sonnet, Gemini Flash, and DeepSeek‑V4‑Pro. DeLM outperforms other models on SWE-Bench for a number of reasons , as Mao detailed on X. First, agents share failures. In ordinary parallel runs, when one agent follows the wrong path, that failure stays private, and subsequent agents may waste time (and money) pursuing the same dead end. But with DeLM, failed hypotheses are written into shared context. “Later agents can read them as constraints, avoid repeated exploration, and redirect their search toward more promising fixes,” Mao said. Additionally, constraints, once verified, are immediately added to agents’ shared context. This means they become a binding shared state. “Later agents inherit them, build around them, and avoid repeating globally invalid simplifications,” Mao said. Crucially, DeLM keeps shared progress compact enough to reuse. It is unfoldable, meaning agents see short gists by default, but can choose to unfold them into more detailed summaries and raw evidence. As the researchers note, providing all raw documents and traces gives agents the maximum amount of information, but that can overwhelm their context windows and ultimately increase costs. “If agents shared full traces, each worker would need to read long command histories, file dumps, failed edits, and intermediate reasoning, turning coordination itself into another long-context bottleneck,” Mao said. On the other hand, while sharing compact summaries is cheaper, important details and evidence can be lost, resulting in less reliable reasoning. Unfolding, therefore, provides “coarse-to-fine” opt-in access. This can improve accuracy and cost. Ultimately, with a framework like DeLM, agents can be more efficient because they are prevented from repeatedly reading the same documents or rerunning the same failed analysis; more effective because useful findings are propagated across parallel threads; and more robust because they only share verified claims. For enterprise builders, DeLM challenges a core assumption: that every multi-agent workflow needs a central controller. The SWE-bench and LongBench-v2 results suggest the decentralized model isn't just theoretically cleaner — it's faster, more accurate, and roughly half the cost.
- Ecosystem Roundup: Vietnam’s AI surge is more than a funding spike
Vietnam’s AI investment surge is striking not just for its scale, but for what it signals about the country’s evolving economic identity. A leap to US$130M in 2025 suggests that Vietnam is no longer merely a manufacturing story or a low-cost outsourcing destination; it is trying to position itself as a serious node in the […] The post Ecosystem Roundup: Vietnam’s AI surge is more than a funding spike appeared first on e27 .
Score: 55🌐 MovesJun 16, 2026https://e27.co/ecosystem-roundup-vietnams-ai-surge-is-more-than-a-funding-spike-20260616/ - Databricks’ new agentic coworker Genie One brings AI automation to every part of the business
Big-data company Databricks Inc. is getting into the agentic artificial intelligence coworker game with the launch of a new tool called Genie One, aimed at helping business teams orchestrate workflows and automate work-related tasks. The arrival of Genie One expands on the company’s existing Genie suite, but goes well beyond its original conversational analytics capabilities. […] The post Databricks’ new agentic coworker Genie One brings AI automation to every part of the business appeared first on SiliconANGLE .
- Ineffable Intelligence strikes Google Cloud deal for Vera Rubin GPU power
Frontier artificial intelligence (AI) startup Ineffable Intelligence has selected Google Cloud as its exclusive infrastructure partner to develop the world’s first “superlearner”. The partnership comes hot on the heels of a $1.1bn (£860m) seed round announcement in April. The deal, announced at the Google Cloud Summit in London this week, will result in the startup deploying one of the world’s largest clusters of Nvidia Vera Rubin NVL72 GPUs (A5X) to power research into systems that can discover knowledge through their own experience rather than human-provided datasets. Ineffable Intelligence was founded by David Silver, the UCL professor and former Google DeepMind scientist who led the AlphaGo and AlphaZero projects. The company is aiming to bypass the “human data ceiling” that it said currently limits large language models (LLMs) such as ChatGPT and Claude. April’s record-setting seed round – the largest in European history – values the London-based company at $5.1bn and includes backing from Sequoia, Lightspeed and the UK government’s Sovereign AI Fund. Unlike traditional AI training which relies on static data, Ineffable’s “superlearner” is designed for experience-based learning where the model generates, evaluates and learns from its own actions in real time. This operational shift places fundamentally different demands on infrastructure and requires the high-performance networking and tightly integrated training systems such as those provided by Google’s AI Hypercomputer architecture. According to David Silver, CEO and founder of Ineffable Intelligence, the decision to partner with Google Cloud was driven by the need for access to orchestrated hardware and software rather than just access to raw compute. “We evaluated the space and chose Google Cloud as the best fit for our reinforcement learning infrastructure,” said Silver. “We aren’t just looking for processors; we are building a resilient and scalable environment to make ‘first contact’ with superintelligence – AI that transcends human limitations in science, mathematics and technology.” The deployment utilises Google Cloud’s full-stack AI Hypercomputer, incorporating Jupiter networking and optimised storage to handle the massive computational scale required for reinforcement learning. This architecture moves away from standard “box of chips” provisioning to provide a systems-level optimisation that ensures researchers can focus on breakthroughs in autonomous learning rather than infrastructure bottlenecks. The project further consolidates London as a critical global centre for frontier AI research, with the startup’s mission expected to attract premier engineering talent to the UK. The backing from the Department for Science, Innovation and Technology (DSIT) and the Sovereign AI Fund reflects a strategic move by the UK government to scale British-built technology that can generate new knowledge in medicine, engineering and science. Industry observers have noted that Ineffable’s “anti-LLM” strategy represents a high-stakes scientific bet. While reinforcement learning proved successful in closed-system game environments like Go and StarCraft , applying trial-and-error algorithms to the vast complexity of human knowledge and scientific discovery remains an unproven frontier. By bypassing human data avoids the inherent flaws and biases of engines designed for mimicry, the “superlearner” path may lack the immediate utility and predictability of current generative AI systems. The technical challenge of ensuring safety and ethical guardrails in a system that discovers knowledge independently of human input will be a big hurdle as the lab attempts to rediscover and then transcend human inventions. In May, Ineffable Intelligence announced collaboration with Nvidia on the engineering requirements for its massive GPU cluster, to ensure the environment can scale to support the next generation of reinforcement learning algorithms. Silver believes this superlearning capability will eventually discover profound intellectual breakthroughs in language and mathematics. The startup was founded in late 2025 and has become a cornerstone of Europe’s AI ecosystem. Read more about AI and Google Cloud Brace for cloud price hikes and AI failures amid pressure to modernise . Organisations risk losing control of their IT infrastructure unless they embrace platform-centric models, modernise procurement and cut through the agentic AI hype, Gartner analysts warn. Google Cloud supplants Azure as Unilever cloud of choice . Microsoft Azure provided ‘the bulk’ of provision when Unilever went all-in on cloud in 2023, but now Google will be the ‘destination’ for the multinational’s cloud and data platform.
- Latest AI News & Market Insights
Latest AI News & Market Insights PitchBook
- Okta expands Google Cloud partnership to secure AI agents and the browser
Okta Inc. today expanded its partnership with Google LLC’s Google Cloud with a set of integrations that bring identity governance to artificial intelligence agents and tighten security across the Chrome browser. The deal pairs Okta’s identity layer with Google Cloud’s Gemini Enterprise Agent Platform and Chrome Enterprise. Enterprises now manage swarms of AI agents the […] The post Okta expands Google Cloud partnership to secure AI agents and the browser appeared first on SiliconANGLE .
Score: 55🌐 MovesJun 16, 2026https://siliconangle.com/2026/06/16/okta-expands-google-cloud-partnership-secure-ai-agents-browser/ - STAT+: Verge Labs’ new AI model solves patient stratification problems for neurology clinical trials
As the saying goes, one man’s trash is another man’s treasure. Or as Verge Labs might put it, one company’s failed clinical trial … is that same company’s new AI…
- Mach Industries wins DIU contract for maritime, long-range strike drone
Mach Industries wins DIU contract for maritime, long-range strike drone Breaking Defense
Score: 55🌐 MovesJun 16, 2026https://breakingdefense.com/2026/06/mach-industries-wins-diu-contract-for-maritime-long-range-strike-drone/