Gate News message, April 28 — Meituan has quietly rolled out a new AI model, LongCat-2.0-Preview, on its LongCat API platform with an update log dated April 20, but has not issued any official announcement or technical report. Unlike previous LongCat series models (Flash-Chat, Flash-Thinking, Flash-Lite, Flash-Omni, Next), which shipped with official blog posts, technical reports, and open-source releases on Hugging Face and GitHub, the 2.0-Preview version offers no open-source links and is available exclusively via API.
The model’s update log highlights three core capabilities: agent development with native support for tool calling, multi-step reasoning, and long-context tasks; proficiency in code generation, workflow automation, and complex instruction execution; and deep integration with Claude Code, OpenClaw, OpenCode, and Kilo Code. According to reports from multiple media outlets citing sources on April 24, the model’s total parameters exceed one trillion, employs a MoE (Mixture of Experts) architecture, and supports a 1 million-token context window—comparable in scale to DeepSeek V4, also released that day.
Insiders revealed that LongCat-2.0-Preview was trained entirely on domestic computing clusters using between 50,000 and 60,000 Chinese-made accelerator cards, marking the largest-scale training task completed on domestic AI infrastructure to date. During the testing phase, the model provides a free daily allowance of 10 million tokens.
Disclaimer: The information on this page may come from third parties and does not represent the views or opinions of Gate. The content displayed on this page is for reference only and does not constitute any financial, investment, or legal advice. Gate does not guarantee the accuracy or completeness of the information and shall not be liable for any losses arising from the use of this information. Virtual asset investments carry high risks and are subject to significant price volatility. You may lose all of your invested principal. Please fully understand the relevant risks and make prudent decisions based on your own financial situation and risk tolerance. For details, please refer to
Disclaimer.
Related Articles
U.S. Semiconductor ETFs Fall in Premarket Trading as AI Stocks Decline Sharply
Gate News message, April 28 — U.S. semiconductor ETFs fell in premarket trading on April 28 as artificial intelligence stocks declined sharply.
iShares Semiconductor ETF dropped 4.2%, while Vaneck Semiconductor ETF fell 3.8%.
GateNews11m ago
Citi raised its AI market valuation to $4.2 trillion and its CapEx forecast to $8.9 trillion
Citi’s April 27 report revised the 2030 global AI market upward to $4.2T, and corporate AI to $1.9T; CapEx for 2026–2030 was also raised to $8.9T. The most notable part is that corporate AI was revised up by 58%, indicating that the adoption pace is far faster than expected. The drivers include AI programming automation and the spread of application-layer CapEx expansion; Anthropic’s 2026 Q1 annualized revenue reached $30B, showing growth. For Taiwan’s technology stocks, CapEx moving higher is a positive catalyst for downstream chip and packaging demand.
ChainNewsAbmedia37m ago
China's MIIT Approves 690 Industry Standards Including AI Deep Synthesis Image System Specification
Gate News message, April 28 — China's Ministry of Industry and Information Technology (MIIT) has approved 690 industry standards, including the "AI Deep Synthesis Image System Technical Specification," according to an official announcement.
The approved standards span multiple sectors: chemical ind
GateNews40m ago
Kimi K2.6 Tops OpenRouter LLM Rankings in First Week, Usage Surges 7,683%
Gate News message, April 28 — Kimi K2.6, the flagship model from Moonshot AI, reached the top of OpenRouter's weekly LLM rankings with 1.88 trillion tokens consumed, surging 7,683% week-over-week. The model, launched on OpenRouter on April 20, surpassed Claude Sonnet 4.6 (1.35T, -3%) and DeepSeek
GateNews1h ago
After the appearance of ChatGPT, submissions to management journals increased by 42%: AI is pushing academia toward “more” rather than “better”
The editorial notes that since ChatGPT went live, the number of posts has increased by 42%, and the editorial department’s staffing has also been significantly expanded. An AI-driven flood of low-quality submissions has doubled the burden of peer review, and the pressure on the initial screening remains particularly acute. Mollick points out that the problem lies in an incentive structure that favors “more” rather than “better.” He argues that mechanisms such as a quality-oriented approach and verifiable research processes are needed to increase credibility, and he warns that other fields requiring human review are facing similar challenges.
ChainNewsAbmedia1h ago
DeepMind × South Korea MOU: First offshore AI campus, Hassabis signs personally
DeepMind CEO Hassabis personally traveled to Seoul to sign an MOU with the South Korean government to accelerate life sciences, climate research, and scientific discovery with AI. The core is to establish an overseas AI Campus at Google Seoul to provide model access and facilitate interaction among research personnel, and to launch a national-level AI science research center in 2026, focusing on life sciences, climate, and basic science. At the same time, it will carry out collaborations with companies such as Samsung, SK Hynix, Hyundai Motor, and LG, demonstrating South Korea’s global competitiveness and influence in AI and semiconductors.
ChainNewsAbmedia1h ago
