AI News

Follow the latest AI and automation news

25 May 20269 Dhuʻl-Hijjah 1447 AH

Alibaba's Qwen 3.7 Max Surprises Overseas Developers with 35-Hour Autonomous Task Run

Alibaba has launched its new model, Qwen 3.7 Max, which has excited international developers by demonstrating an exceptional ability to perform tasks autonomously for 35 continuous hours with 1,158 tool invocations without any performance degradation. The model gained attention after a user on platform X noted its capability to handle long-term tasks independently. This feature positions Qwen 3.7 Max as a leader in managing complex workflows. Its release strengthens Alibaba's position in the large language model race, as developers begin exploring its capabilities in areas such as programming and agent coordination.

Business

25 May 20269 Dhuʻl-Hijjah 1447 AH

How Xiaomi’s push into AI, chips and EVs is future-proofing its hardware empire

Xiaomi is heavily investing in artificial intelligence to ensure it remains competitive in the smartphone and electric vehicle markets. The company recently introduced its new model, MiMo-V2.5-Pro, which was ranked as the top open-source model for agentic capabilities by Artificial Analysis. This move is part of Xiaomi's broader strategy to integrate AI across its product lines. By focusing on open-source models, the company aims to enhance its competitive edge in the rapidly evolving AI landscape.

Business

25 May 20269 Dhuʻl-Hijjah 1447 AH

OpenAI, Grupo Folha and Grupo UOL announce strategic content partnership

OpenAI has partnered with Grupo Folha and Grupo UOL to deliver trusted Brazilian journalism through ChatGPT. This collaboration aims to enhance news access with a focus on attribution and transparency. The initiative allows users to receive accurate information from reputable news sources, thereby boosting content credibility. It also aims to improve user experience by providing reliable and diverse news coverage.

💡

DeepSeek's Price Cut for V4-Pro Impacts AI Model Market

How will the price cut impact competition in the AI market?

Models

ByteDance study finds that asking LMMs questions beats making it transcribe text for long document training

24 May 20268 Dhuʻl-Hijjah 1447 AH

Models

24 May 20268 Dhuʻl-Hijjah 1447 AH

ByteDance study finds that asking LMMs questions beats making it transcribe text for long document training

The ByteDance Seed model, featuring 7 billion parameters, has demonstrated superior reliability in answering questions on long, image-heavy documents compared to larger models. It was tested on documents four times longer than its training data, showcasing its effectiveness in handling complex information. Instead of transcribing text, the model learns by answering questions and identifying relevant passages independently. This approach marks a shift in how large language models are trained, emphasizing comprehension over rote repetition.

Models

Deepmind's Hassabis sees humanity "in the foothills of the singularity" while LeCun says current AI isn't intelligent

24 May 20268 Dhuʻl-Hijjah 1447 AH

Models

24 May 20268 Dhuʻl-Hijjah 1447 AH

Deepmind's Hassabis sees humanity "in the foothills of the singularity" while LeCun says current AI isn't intelligent

Yan LeCun believes that current AI systems are not truly intelligent, lacking the ability to learn from experience. In contrast, Demis Hassabis sees humanity at the threshold of a transformative shift toward general artificial intelligence. Oriol Vinyals adds that models today would have seemed like general AI seven years ago, yet they still need to achieve real breakthroughs. The differing opinions among these AI pioneers address the challenges facing current systems. LeCun emphasizes the absence of genuine intelligence, while Hassabis discusses the potential future advancements.

Business

DeepSeek V4 Pro tops global bang-for-buck ranking after 75% price cut

24 May 20268 Dhuʻl-Hijjah 1447 AH

Business

24 May 20268 Dhuʻl-Hijjah 1447 AH

DeepSeek V4 Pro tops global bang-for-buck ranking after 75% price cut

Chinese AI start-up DeepSeek has announced a permanent 75% price cut on its V4 Pro model, making it one of the world's top performers in intelligence-per-dollar efficiency. This announcement follows the release of the much-anticipated V4 generation, which includes the flagship V4 Pro. This strategic move positions DeepSeek favorably against US competitors like OpenAI and Anthropic, highlighting its cost-effectiveness. The significant price reduction could reshape competitive dynamics in the global AI market.

Agents

Anthropic may keep supplying Claude to the NSA despite being flagged as a supply chain risk by the Pentagon

24 May 20268 Dhuʻl-Hijjah 1447 AH

Agents

24 May 20268 Dhuʻl-Hijjah 1447 AH

Anthropic may keep supplying Claude to the NSA despite being flagged as a supply chain risk by the Pentagon

Anthropic is likely to continue supplying AI models to the NSA, despite being labeled a supply chain risk. Intelligence agencies are lacking Nvidia's latest Grace Blackwell chips, and Anthropic's Mythos model reportedly operates on older hardware. The controversial 'any lawful use' clause that previously derailed negotiations is not part of the current deal, facilitating ongoing collaboration between Anthropic and the NSA. This development comes as agencies face shortages of advanced technology needed for their operations.

Models

Researchers let Claude Code discover AI scaling algorithms that humans probably wouldn't have designed

24 May 20268 Dhuʻl-Hijjah 1447 AH

Models

24 May 20268 Dhuʻl-Hijjah 1447 AH

Researchers let Claude Code discover AI scaling algorithms that humans probably wouldn't have designed

Researchers from the University of Maryland, Google, and Meta utilized AutoTTS technology to enable a software agent to discover new control algorithms. The discovered algorithm achieved a 70% reduction in computing costs compared to traditional methods while maintaining accuracy. The research took only 160 minutes and cost a total of $40, demonstrating the effectiveness of this technology. This advancement is significant in artificial intelligence, allowing systems to enhance their performance independently without human intervention.

Agents

AgentX AI Hackathon Kicks off to Empower National Talent | Ministry of Communications and Information Technology

24 May 20268 Dhuʻl-Hijjah 1447 AH

Agents

24 May 20268 Dhuʻl-Hijjah 1447 AH

AgentX AI Hackathon Kicks off to Empower National Talent | Ministry of Communications and Information Technology

The AgentX AI Hackathon was officially launched on October 18, 2025, in partnership with the Ministry of Communications and Information Technology. This initiative aims to empower national talent to create real-world AI solutions addressing challenges in vital sectors. Building on ongoing efforts within the digital ecosystem, the hackathon nurtures technological human capital and strengthens collaboration between Saudi talent and the business community. It provides a competitive platform for participants to apply their skills and foster innovation.