Major Breakthrough in Domestic Models! Kuaishou Unveils High-Performance Agent Model SkyClaw-v1.0

Kunlun Wanyi Group officially launched the high-performance Agent model SkyClaw-v1.0, tailored for real-world workflows, and simultaneously released the lightweight version SkyClaw-v1.0-lite, which balances performance and cost-effectiveness. The launch of this model marks a solid step forward for domestic large models in the construction of intelligent agent ecosystems and long-text processing capabilities.

Supports Millions of Context Tokens and Multi-Step Task Execution

SkyClaw-v1.0 features an ultra-long context capacity of up to one million tokens, and has been deeply optimized for real intelligent agent scenarios such as complex tool calls, multi-turn task execution, code generation, and interactive application development. The model is now natively integrated into the TianGong Skywork platform and can run smoothly in mainstream Agent environments such as OpenClaw, Hermes, and Nanobot.

In mainstream Agent evaluations, SkyClaw-v1.0 demonstrated highly stable multi-step task execution capabilities. Its overall performance not only comprehensively surpasses mainstream open-source models like Minimax 2.7 and DeepSeek V4 Flash, but also approaches top-tier models such as DeepSeek V4 Pro and Claude Opus 4.6 in specific tasks.

High Cost-Effectiveness and Time-Limited Free Trial

In addition to technological breakthroughs, SkyClaw-v1.0 also shows great sincerity in commercialization. The model's market pricing is highly competitive, directly below half of the price of Minimax 2.7 and Qwen 3.6 series models, laying a solid cost foundation for the large-scale utilization of high-performance Agent capabilities.

To allow developers and enterprise users to experience its outstanding performance in advance, Kunlun Wanyi announced that it will officially open a free trial period lasting 2 to 4 weeks starting today. Users just need to log in to the official website and open TianGong Skywork to experience its full set of core capabilities, including long-context understanding, tool calling, and research-level data analysis, directly in real tasks without configuration.

Google Cracks Down on AI Poisoning: GEO Spam Content Will Be Devalued and Removed

Google updated its generated search governance guidelines in mid-May, officially classifying malicious manipulation behaviors in Generated Engine Optimization (GEO) as spam content. This move targets the challenge of "AI poisoning" in the era of large models, aiming to maintain the fairness of the search ecosystem by taking a stand against advertisers attempting to deceive "AI" by fabricating information.

Global AI Sector Financing Exceeds 110 Billion Yuan in the First Quarter, Domestic Large Models Receive Over 30 Billion Yuan in Investments in May

The artificial intelligence venture capital market has been heating up recently. In the first quarter, there were nearly 600 financing activities in the global AI sector, with a total amount exceeding 110 billion yuan, representing a year-on-year increase of 185.4%. In May, the domestic market remained strong, with large model companies such as Moonshot and Stepwise Star receiving over 30 billion yuan in financing. The embodied intelligence sector also gained popularity, with startups such as Vite Dynamics and Lu Ming Robotics securing several billion yuan in financing within a week. Funds are flowing quickly into the market, and the enthusiasm remains high.

Q1 AI Investment Booms! Domestic Large Models See Funding Surges to 30 Billion Yuan!

In the first quarter of this year, there were nearly 600 AI-related financing events with a total amount exceeding 110 billion yuan, a year-on-year increase of 185.4%. Domestic large model projects such as "Moonlight Darkness" and "Step Star" secured over 30 billion yuan in funding in May, and the industry trend continues to heat up.

The First Stock of Global Large Models Lands in Beijing: Zhitu AI Acquires a 22,700 Square Meter Headquarters Building for 361 Million Yuan

Zhitu AI completed the acquisition of 100% ownership of Beijing Hongzuan Technology with a total cost of no more than 361 million yuan (including 81.62 million yuan for equity acquisition and 278.9 million yuan for debt assumption), thereby obtaining the property of the Diamond Building in Zhongguancun Software Park. The building is located on Dongbeiwang West Road in Haidian District, with a construction area of about 22,700 square meters. It is surrounded by technology companies such as Baidu, Tencent, and Lenovo, marking the official establishment of Zhitu AI's headquarters property.

Million-Dollar Seed Round Completed, Can the MemoraX AI Memory Model End AI Amnesia?

Hao Jianye believes that if AI cannot move from 'storage' to 'memory,' it cannot become a true intelligent partner. The company he founded, MemoraX AI, completed a million-dollar seed round just one month after its establishment, led by the L2F Lightsource Entrepreneur Fund and Zhongding Capital. The funds will mainly be used for research and development of Agentic RL (Agent Reinforcement Learning) technology, aiming to break through the limitation of large models being merely efficient search engines.