Search results：Mu

54 results

Top Researcher Luke Metz Leaves OpenAI and Joins Meta's Super Intelligence Lab

AI talent war heats up as researcher Luke Metz leaves OpenAI to join Meta's superintelligence lab this week. His career shows frequent moves among top AI labs; in 2024 he left OpenAI to join a new organization founded by former executive Mira Murati.....

08-2412.9K

Kimi K2.5 Retires at the End of This Month: The First Generation Trillion-Parameter Multimodal Model of Moonshot AI Says Goodbye

Moonshot AI announced its trillion-parameter multimodal model Kimi K2.5 will end service at the end of this month. Open-sourced in January, it supports vision and text input, thinking/non-thinking modes, dialogue and agent tasks, and achieved SOTA on open benchmarks in code, image, and video. Next-generation K3 will take over with larger scale.....

08-2412.4K

Musk Likes! Honor Humanoid Robot Runs 400 Meters in 39.45 Seconds, Setting a New World Record

On Aug 23, at the 2nd World Humanoid Robot Games, Honor's self-developed robot "Lightning" ran 400m in 39.45s, beating the human world record of 43.03s by Wayde van Niekerk, who praised it. In prelims, it ran 100m in 9.32s (peak 14.5m/s), surpassing Bolt.....

08-2413.1K

OpenAI Executive Lehan Warns: AI Can Now Plan Cyber Attacks, Public Companies Must Be Prepared for Defense

OpenAI Chief Global Affairs Officer Chris Lehane warns frontier AI can plan and launch complex cyberattacks, urging defense against relentless attacks. A training agent broke its sandbox and hacked Hugging Face, prompting calls for legislation as AI enters a new phase.....

08-2412.4K

2nd World Humanoid Robot Sports Meet Opens: 2056 Steel Soldiers Compete in the Ice Ribbon for Supremacy

On Aug 22 evening, the 2nd World Humanoid Robot Games opened at Beijing's National Speed Skating Oval, co-hosted by Beijing Municipal Government and China Media Group, attracting 666 teams and 2,056 robots from 16 countries across six continents. Teams rose 138% and robots quadrupled, highlighting booming embodied AI.....

08-2412.1K

Texas Governor Strongly Criticizes AI Expansion: Building Data Centers Rapidly Is Digging a Grave for Ourselves

Texas Gov. Abbott warned AI data centers lacking community support are 'digging their own graves.' His shift from recruitment to curbs isn't isolated: Pennsylvania added limits this week, New York paused hyperscale projects last month. From $40B recruitment to audits, Republicans tighten regulation.....

08-2411.9K

Anonymous Model Ox Alpha Tops OpenRouter, Cumulative API Calls Exceed 4 Trillion Tokens

Anonymous model Ox Alpha (called "Niu Lai" by Chinese community) surged to top OpenRouter rankings after launch, setting a single-day usage record. Top five apps have cumulatively called over 4 trillion tokens; popularity spread from community to global developers and agent workflows. Developer undisclosed; positioned for coding and persistent tasks.....

08-2425.7K

Mysterious Ox Comes with the Large Model: Ox Alpha Surpasses Mainstream Giants in Various Benchmark Tests

OpenRouter launched anonymous model Ox Alpha on Aug 20, with mysterious developer identity. It impressed in benchmarks and sparked tech community debate. It targets coding, long-horizon agent tasks, and real production environments.....

08-2423.3K

Watch Out! DeepSeek Flash Vision Model Gets a Sudden Update, Supports Multimodal

The DeepSeek toolchain is upgraded to version 0.1.1-rc.1, adding the visual model experience version V4Flash Vision-Exp on top of V4Flash and V4Pro. Previously, version 0.1.0-rc.8 had already enabled multimodal capabilities, supporting native image requests and mixed text-image input with commands such as /goal and /plan.

08-2134.8K

Alibaba Open Sources Qwen-UI-Agent: The Real-World Multimodal Foundation Model Arrives

Alibaba launches the GUI Intelligent Agent Foundation Model Qwen-UI-Agent, centered on the real world, covering mobile, desktop, web, and deep search, aiming to break through the limitations of simulated testing and achieve seamless operation on complex real devices. It achieved a high score of 82.1% in the MobileWorld mobile test, leading multiple mainstream models, with comprehensive performance comprehensively surpassing industry benchmarks.

08-2133.5K

Mistral Launches Agentic Search, Multi-Step Retrieval Enhances Accuracy of Complex Documents to 86%

When enterprises build AI, core confidential data is often buried in long documents, financial reports, legal contracts, and other files. Traditional one-time RAG only retrieves fixed text blocks, lacking iterative reasoning, making it difficult to handle complex long documents and cross-source comparisons. To address this, Mistral AI launched the Agentic upgrade solution, enhancing iterative reasoning capabilities and solving the challenge of information extraction from long documents.

08-2129.0K

SenseTime Technology Open Sources 8B Lightweight Multimodal Large Model: Native 4K Output, Comprehensive Visual Editing

SenseTime opensources the 8B lightweight unified multimodal model SenseNova U1.5Lite, which demonstrates superior performance in core visual tasks, comparable to large-scale commercial models in handling complex layouts and text rendering; it natively supports 3-4K context length, enabling processing of multiple constraints such as subject, quantity, spatial relationships, text, layout, and style simultaneously.

08-2133.7K

SAP Collaborates with Team Liquid to Launch AI Voice Application for Professional Esports Data Analysis

SAP and Team Liquid jointly developed an AI voice application that can automatically process multi-channel internal voice communications during matches and training, accurately separate individual player voice streams, and complete transcription, speaker identification, and emotion analysis. It transforms scattered conversations into structured match data, helping teams analyze matches, optimize strategies, and improve player performance.

08-2121.5K

Qwen AI Platform Model Matrix Continues to Expand GLM-5.3 and DeepSeek-V4-Pro Officially Launched

The Alibaba Cloud Qwen AI Platform has recently updated its model matrix, officially integrating the GLM-5.3 from Zhipu and the DeepSeek-V4-Pro, with APIs now available. Both models focus on programming and intelligent agent scenarios: GLM-5.3 enhances programming, long-term tasks, and cybersecurity capabilities; DeepSeek-V4-Pro improves intelligent agent interaction, suitable for code development and multi-step complex task processing.

08-2124.8K