News reporting focused on AI and machine learning, covering the companies behind these technologies, their real-world applications, and the ethical concerns they raise. This includes areas like generative AI (large language models, text-to-image and video), speech tech, and predictive analytics.
DeepSeek has announced general availability of V4-Pro and beta of V4-Flash, with API price increases up to 1,100% for the V4 family, effective August 16. The hikes, driven by capacity constraints, introduce peak and off-peak pricing. Analysts note the increases erode DeepSeek's price advantage over OpenAI at peak times, but off-peak rates and cache discounts still offer savings for many buyers.
Leaked minutes from a four-hour meeting reveal DeepSeek CEO Liang Wenfeng's singular focus on learning as the path to AGI, his open-source business strategy, and his management philosophy. The document offers rare insight into the mind of a key figure in the US-China AI race, covering topics from API revenue to domestic chip reliance.
Nvidia is developing Nemotron 4, an open-weight AI model family with a trillion-parameter flagship, expected as early as fall 2026. However, Chinese labs like Moonshot AI and DeepSeek already ship larger models, with Kimi K3 at 2.8 trillion parameters and DeepSeek V4 Pro at 1.6 trillion. Nvidia has tripled cloud spending to $28 billion through 2031, betting on open models to drive GPU demand, even as it competes with customers like OpenAI.
Microsoft's new MAI Code 1.1 Flash model for GitHub Copilot underperforms DeepSeek's open-weight alternative on Terminal Bench 2.1 while costing more per token. Independent benchmarks show DeepSeek-V4-Flash-0731 scoring 82.7% versus Microsoft's 62.9%, despite Microsoft's claims of improved efficiency. The gap raises questions about Microsoft's open AI positioning as it pushes proprietary models.
Composio benchmarked DeepSeek V4 Flash across four agent frameworks, finding no clear winner. Oh My Pi led in success rate (17/30), Claude Code was fastest but priciest at $0.195 per task, and OpenCode was cheapest at $0.073. Cost varied nearly 3x and speed 2.2x, with seven tasks succeeding or failing solely based on framework choice.
The latest wave of open-weight model releases shows a field far more crowded and dynamic than predicted, with new entrants like Thinking Machines and Tencent making strategic moves. DeepSeek, Moonshot, and Poolside push boundaries, while hardware milestones like Meituan LongCat's training on Ascend 910s mark significant progress. The article revisits the consolidation question, noting that more companies are training strong models and releasing them openly, challenging earlier predictions. With license shifts and global lab entries, the open model movement is expanding in scope and ambition.
A Chinese threat actor deployed a DeepSeek-powered Hermes Agent that autonomously hunted for vulnerable servers, selected targets, downloaded exploits, and changed tactics when it failed. Authentication stopped the agent before it compromised any targets, according to Palo Alto Networks' Unit 42. The incident marks one of the first documented cases of an AI agent carrying out a full attack chain without human oversight, exposing the operator's infrastructure in the process.
A Chinese threat actor deployed a DeepSeek-powered Hermes Agent that autonomously targeted and attacked vulnerable servers, according to Unit 42. Authentication stopped the agent before compromise, but the incident reveals the limits of Zero Trust against agentic AI. Experts warn that such attacks will scale faster than defenses, requiring new approaches to authority and verification.
Researchers introduce Autonomous Topology Mutation (ATM), a runtime team-mutation mechanism for multi-agent LLM frameworks. ATM uses telemetry-driven overload detection and three safety invariants to restructure agent teams without downtime. On 720 DeepSeek-V3-driven task runs, ATM lifted code-task success from 3.3% to 61.7% while eliminating high-privacy memory exposure.
New research from Yuhang Wang introduces PlanFlip, a framework of four prompt injection attacks targeting the planning phase of multi-agent LLM systems. The attacks exploit a single injection into the Planner agent's context to corrupt all downstream sub-tasks. Testing on nine frontier LLMs across 3,479 episodes revealed that stronger models like GPT-5 are more vulnerable, while reasoning-augmented models like DeepSeek-R1 show full resistance.
Chinese company Moonshot AI released Kimi K3, an open source model that rivals top proprietary models. The launch triggered a Nasdaq dip and renewed debate among tech figures about Chinese AI progress, distillation, and national security risks, echoing the DeepSeek controversy from early 2025.
Deepseek topped Ramp's list of fastest-growing software vendors in June 2026, driven by US companies seeking cheaper AI. Ramp's chief economist Ara Kharazian warned about security risks of using Chinese models directly and questioned the trend's durability. Deepseek's V4 model costs a fraction of Western rivals but trails in overall performance.