Meituan releases weights for LongCat-2.5, a 1.6T-parameter MoE model for long-horizon agentic tasks, under the MIT license, after unveiling it on November 1 (Carl Franzen/VentureBeat)
Developers now have unrestricted access to a massive system specialized in handling extended, multi-step automated workflows. Distributing these core assets under a permissive open-source license allows the broader community to study and implement the underlying architecture directly. This availability marks a major transition from the system's earlier demonstration to active public deployment. Specifically, the release provides open access to the full weights following the model's public debut on November 1.
Key points
The system is explicitly tailored to process and execute long-horizon agentic tasks.: The system is explicitly tailored to process and execute long-horizon agentic tasks.
Broad access to the architecture is granted through the permissive terms of the MIT licen…: Broad access to the architecture is granted through the permissive terms of the MIT license.
The availability of the model weights builds directly on the earlier November 1 unveiling.: The availability of the model weights builds directly on the earlier November 1 unveiling.
Sources checked by our analysis pipeline: 1
Associated Press: |
New York Times: |
Semafor: |
Sarah Perez / TechCrunch: |
Lauren Goode / Wired: |
Mackenzie Hawkins / Bloomberg: |
Demetri Sevastopulo / Financial Times: |
Financial Times: |
Mark Anderson / Bloomberg: |
Anthropic: |
Jay Peters / The Verge: |
Ryan Christoffel / 9to5Mac: |
Russell Brandom / TechCrunch: |
Anna Tong / Forbes: |
Reuters: |
Echo Wang / Reuters: |
Ryan Lawler / Axios: |
Emily Nicolle / Bloomberg: |
Jason Schreier / Bloomberg: |
Elias Schisgall / Wall Street Journal: |
Reuters: |
Carl Franzen / VentureBeat: |
Jenny Lee / CNBC: |
Associated Press: |
Kanishka Singh / Reuters: |
This is a Techmeme archive page. It shows how the site appeared at 3:50 AM ET, November 10, 2026.
The most current version of the site as always is available at our home page. To view an earlier snapshot click here and then modify the date indicated.
With hundreds of millions of users turning to chatbots for medical guidance, experts warn that technically correct information can still trigger harmful public health decisions. Current government evaluations focus heavily on truthfulness, overlooking how subtle messaging cues can discourage vital actions like cancer screenings and immunizations. To prevent long-term harm, specialists are urging regulators to measure how conversational models directly influence human behavior rather than relying solely on fact-che…
Congressional lawmakers are leveraging personal controversy around Nvidia's leadership to push for stricter legislative controls over foreign artificial intelligence hardware distribution. The bipartisan effort aims to integrate tighter monitoring directly into defense authorization laws amid fears that foreign entities are bypassing tech restrictions through intermediary Asian facilities. U.S. officials remain focused on potential backdoor routes that allow restricted processing power to flow toward Beijing.
Tech leaders are revamping their automated infrastructure as complex user assistance tasks expose the limits of conventional automation. Airbnb Chief Technology Officer Ahmad Al-Dahle revealed that the travel company is turning to multi-party AI agents to tackle these obstacles. This architectural shift requires continuous, rigorous testing to prevent service slowdowns while ensuring dependable, precise assistance across user touchpoints.