DeepSeek and MiniMax have successfully fulfilled registration procedures, completing open-source and overseas licensing. Most artificial intelligence technologies can still be freely exported, and overseas localized deployment falls under the scope of technology export management.
Prime Intellect Blog3838Prime Mega MoE enables the DeepSeek Mega MoE kernel to be used for training by integrating backpropagation kernels, achieving up to 3.11 times faster MoE kernel speed compared to the default PyTorch implementation, and 1.34 times faster end-to-end training speed compared to the Prime-RL baseline.
DeepSeek and MiniMax, two leading companies, have successfully completed the registration procedures, fulfilling open-source and overseas authorization use. The revised 'China Prohibited and Restricted Export Technology Catalog' by the Ministry of Commerce involves 23 AI-related technologies, raising questions from the outside about a policy shift. This open-source and authorization use indicate that the core of the 'Catalog' is to define boundaries rather than simply restricting the export of AI technologies.
DeepSeek has open-sourced DeepInfer, the foundational framework supporting its large-scale reasoning service, along with the accompanying token codec component deepseek-recipe, under the Apache 2.0 license. DeepInfer is the core of DeepSeek's reasoning service; after the framework is open-sourced, users can achieve an AI gross margin of 82%.
MarkTechPostSame story · 3 outletsDeepSeek has open-sourced the DeepInfer inference framework, which is used to run its production services and support DeepSeek-V4.1-Flash.
SemiAnalysisDeepSeek released the Ascend support for TileLang, DeepGEMM, DeepEP, TileKernels, FlashMLA, and DeepSelect before the National Day holiday. Huawei provides hardware and native software stack, while DeepSeek contributes workload and engineering capabilities, complementing each other to optimize model performance.
The Decoder6969The U.S. Department of Homeland Security has banned the use of AI technology from Chinese laboratory Deepseek by federal civilian agencies, requiring the removal of existing products and prohibiting the purchase of new products, covering its affiliated companies and third-party services, but no specific timeline or specific security concerns have been set. The ban is based on the 2018 U.S. Federal Acquisition Supply Chain Security Act, but it does not explicitly state the related security issues or the time when agencies need to complete the removal.
The DecoderSame story · 3 outlets7272DeepSeek open-sources its internal reasoning framework DeepInfer under the Apache License 2.0, a service software that supports the large-scale and fast answer generation of the V4.1-Flash model.
Liquid AI BlogLFM2.5-24B-A2B is a 24 billion parameter reasoning model that uses only 2 billion active parameters, suitable for fast intelligent agents on devices. The model competes with Qwen3.6-35B-A3B in five key benchmark tests, 1.7 times faster than similar models, and is open source.
The Decoder7070ARTEX, an open-source penetration testing tool using DeepSeek v4.1-flash, GLM-5.3, and Grok 4.6 AI models, was suspected of being used by Chinese attackers to attack multiple South Korean financial institutions from late September to early October 2026, stealing large amounts of data, including over 25,000 records from Shinhan Bank. This incident highlights the risk of AI tools exacerbating the threat of large-scale single-point cyber attacks.
LangChain Deep Agents releases1212LangChain Deep Agents 0.1.80 released, fixing a bug that kept the session record position when opening a selector, and improving skill path handling and cache expiration notification features.
DeepSeek · Hugging Face new modelsDeepSeek-V4.1-Flash is a 552B parameter multimodal MoE model that supports context up to one million tokens, natively processing images and text and generating text autoregressively. It adopts a Causal Encoder-Decoder architecture, reducing global KV cache usage to 890 bytes/token through technologies like CSA2 and FP4 primary KV cache, a 75% reduction compared to DeepSeek-V4-Flash.
Prime Intellect BlogDeepSeek releases SYNTHETIC-2, which includes 20 complex reinforcement learning tasks and verification reasoning trajectories, supporting heterogeneous computing from consumer-grade GPUs to super-large-scale clusters, using the DeepSeek-R1-0528 model to generate data and label difficulty.
Dify Blog6565Dify's low-code platform combined with DeepSeek R1 can quickly build enterprise-level multilingual document translation tools. Through a visual workflow, it achieves document parsing, stylized translation, and batch processing, enhancing cross-language collaboration efficiency without complex programming.
Dify BlogDeepSeek R1 performed excellently in the AIME Mathematics Competition, with accuracy increasing from 15.6% to 71.0%, and reached a human level of 96.3% in the Codeforces Programming Competition, demonstrating strong reasoning ability.
Dify BlogDify combined with DeepSeek can deploy private AI assistants and build local DeepSeek R1 + web search applications. Dify supports deploying the DeepSeek R1:7B model (175B parameters), integrating web search and document analysis, providing support for 1000+ models, ensuring data is completely private. Manage DeepSeek models via Ollama and create chatbots or workflow applications on the Dify platform.
