DeepSeek在其官宣发布DeepSeek-V3.1的文章中提到，DeepSeek-V3.1使用了UE8M0 FP8 Scale的参数精度。另外，V3.1对分词器及chat template进行了较大调整，与DeepSeek-V3存在明显差异。DeepSeekg官方在置顶留言里表示，UE8M0 FP8是针对即将发布的下一代国产芯片设计。 https://t.co/ydxMxF53VL外汇交易员 (@myfxtrader):DeepSeek刚刚官宣V3.1模型。模型同时支持思考模式与非思考模式；思考效率提升，相比DeepSeek-R1-0528，DeepSeek-V3.1-Think能在更短时间内给出答案；Agent能力更强，通过Post-Training优化，新模型在工具使用与智能体任务中的表现有较大提升。 https://t.co/ajVjLgC4ri ...

Core Insights - The 2025 Fortune China Tech 50 list was released, with Huawei Investment Holding Co., Ltd. ranking first due to its comprehensive leadership in communication, chips, operating systems, and artificial intelligence [1] - DeepSeek, focused on AI large model development, ranked second, showcasing significant advancements in the AI sector [1] Huawei - Huawei is recognized as a representative of Chinese tech companies, driving global communication and smart technology development [3] - In the 5G communication sector, Huawei holds a 15% share of essential patents, ranking first globally and providing key technological support for 5G network construction in multiple countries [3] - The Kirin 9020 chip has over 90% localization rate, with significant performance improvements; the Ascend AI processor has set global performance records in supporting large model inference [3] - The Harmony operating system has surpassed 10 million devices, with its ecosystem continuously expanding [3] - Huawei has made breakthroughs in optical technology, network architecture, and AI algorithms, facilitating technological upgrades across various industries [3] DeepSeek - DeepSeek is a leading enterprise in China's AI field, gaining global attention with its self-developed large model DeepSeek-R1, which scored 88.5 in the MMLU benchmark test, significantly outperforming international models like Llama 3 and Claude 2 [9] - The influence of DeepSeek in the open-source community is growing, with its model's global download volume ranking in the top ten [9] - As of June 2025, DeepSeek has reached 163 million monthly active users, becoming the largest AIGC application globally, reflecting China's strong market vitality and technological implementation capabilities in AI [9] Other Notable Companies - Other companies such as CATL, Alibaba, Tencent, and BYD have also made it to the top ranks due to their leading technological capabilities and market performance [10] - CATL continues to lead globally in power batteries and energy storage systems, achieving a 37.6% market share in global power battery installations in 2024, marking its seventh consecutive year in the top position [10]

Core Insights - DeepSeek officially released DeepSeek-V3.1 on August 21, featuring significant upgrades in its architecture and capabilities [1] Group 1: Product Upgrades - The new hybrid reasoning architecture allows a single model to support both thinking and non-thinking modes [1] - DeepSeek-V3.1-Think demonstrates improved efficiency, providing answers in a shorter time compared to DeepSeek-R1-0528 [1] - Enhanced agent capabilities have been achieved through post-training optimization, resulting in better performance in tool usage and agent tasks [1] Group 2: API and Pricing Changes - The DeepSeek API has been upgraded to include deepseek-chat for non-thinking mode and deepseek-reasoner for thinking mode, with context expanded to 128K [1] - A new pricing structure for API calls will be implemented starting September 6, 2025, with the cancellation of night-time discounts [2] - Until September 6, all API services will continue to be billed under the original pricing policy [4] Group 3: Technical Specifications - DeepSeek-V3.1 utilizes UE8M0 FP8 Scale parameter precision, designed for the upcoming generation of domestic chips [4] - The integration of DeepSeek-V3.1 capabilities into the Claude Code framework is facilitated by support for Anthropic API format [1]

Core Insights - DeepSeek has released version 3.1, marking a significant step towards the "Agent Era" [1] - The new version utilizes UE8M0 FP8 Scale parameter precision, indicating advancements in technology [1] - There are notable adjustments in the tokenizer and chat template compared to the previous version, DeepSeek-V3 [1] - The UE8M0 FP8 is specifically designed for an upcoming next-generation domestic chip [1][2] Company Developments - The official webpage, app, mini-program, and API platform have all been updated to incorporate the new model [2] - Users have expressed anticipation for additional features, such as image and video functionality [2]

凤凰网科技讯 8月21日，DeepSeek在其官宣"正式发布DeepSeek-V3.1"的文章里面提到，DeepSeek-V3.1 使用了UE8M0 FP8 Scale的参数精度。另外，V3.1对分词器及chat template进行了较大调整，与 DeepSeek-V3 存在明显差异。DeepSeek官微在置顶留言里说，UE8M0 FP8是针对即将发布的下一代国产 芯片设计。 此外，针对网友提问DeepSeek版本信息不是V3.1的问题，官方回复表示，当前官方网页端、App、小程 序及 API 开放平台所调用模型均已同步更新，新模型自我认知为DeepSeek-V3。 ...

"迈向智能体时代的第一步"。 8月21日，业界千呼万唤的R2模型没来，但DeepSeek官方正式发布了新模型V3.1。从命名来看这或许不是一次大的版本更新，更像是前一代DeepSeek-V3模 型的小版本迭代。 在X上，DeepSeek将V3.1称为"我们迈向智能体时代的第一步"（our first step toward the agent era）。本次升级主要有三大亮点，其中包括更强的 Agent能力、 混合思考模式和更高的思考效率。 官方表示，通过后训练优化，新模型在工具使用与智能体任务中的表现有较大提升。在编程智能体、搜索智能体测评中， V3.1 相比之前的 DeepSeek 系列 模型都有明显提高。 | Benchmarks | DeepSeek-V3.1 | | --- | --- | | SWE-bench | 66.0 | | Verified | | | SWE-bench | 54.5 | | Multilingual | | | Terminal-Bench | 31.3 | DeepSeek-V3.1 是混合推理架构，一个模型同时支持思考模式和非思考模式。目前用户可在官方 App与网 ...

据DeepSeek官方公众号消息，DeepSeek-V3.1正式发布。 本次升级包含以下主要变化： 官方App与网页端模型已同步升级为DeepSeek-V3.1。用户可以通过"深度思考"按钮，实现思考模式与非思考模式的自由切换。 价格调整 将于北京时间2025年9月6日凌晨起，对DeepSeek开放平台API接口调用价格进行如下调整： 执行新版价格表（如下图所示，详见定价页面）； 取消夜间时段优惠。 在9月6日前，所有API服务仍按原价格政策计费，可继续享受当前优惠。 混合推理架构： 一个模型同时支持思考模式与非思考模式； 更高的思考效率： 相比DeepSeek-R1-0528，DeepSeek-V3.1-Think能在更短时间内给出答案； 更强的Agent能力： 通过Post-Training优化，新模型在工具使用与智能体任务中的表现有较大提升。 下载财联社APP获取更多资讯 准确 快速 权威 专业 7x24h电报 头条新闻 VIP资讯 实时盯盘 ...

Core Insights - DeepSeek has officially released DeepSeek-V3.1, introducing a hybrid reasoning architecture that allows users to switch between "Deep Thinking" mode and "Non-Thinking" mode for enhanced interaction [2][3]. Group 1: Hybrid Reasoning Architecture - The "Deep Thinking" mode (DeepSeek-Reasoner) is designed for tasks requiring deep reasoning, such as mathematical calculations and complex logic analysis, providing higher reasoning efficiency [3]. - The "Non-Thinking" mode (DeepSeek-Chat) is tailored for everyday conversations and information queries, offering quicker responses [4]. - Users can easily switch modes via a "Deep Thinking" button on the official app and web interface, enhancing the user experience [5]. Group 2: Enhanced Agent Capabilities - DeepSeek-V3.1 has significantly improved tool usage and agent task performance through Post-Training optimization, resulting in fewer required iterations and higher efficiency in code repair and command line tasks [6]. - Benchmark results show that DeepSeek-V3.1 outperforms its predecessor, DeepSeek-R1-0528, in various tasks, including SWE-bench and Terminal-Bench, with scores of 66.0 and 31.3 respectively [7][8]. Group 3: Efficiency Improvements - The new version employs a thought chain compression training method, reducing output tokens by 20%-50% while maintaining performance levels comparable to DeepSeek-R1-0528, leading to faster response times and lower API call costs [9]. Group 4: API Upgrades and Model Availability - The DeepSeek API has been upgraded to support a context length of 128K, facilitating easier handling of long documents [10][12]. - The base and post-training models of DeepSeek-V3.1 are now open-sourced on platforms like Hugging Face and ModelScope, with a price adjustment for the API set to take effect on September 6, 2025 [11].

Core Viewpoint - DeepSeek has officially released version V3.1, featuring significant upgrades in reasoning architecture, efficiency, and agent capabilities [3][4]. Group 1: Key Features of DeepSeek-V3.1 - The new hybrid reasoning architecture allows the model to support both thinking and non-thinking modes simultaneously [3]. - Enhanced thinking efficiency enables DeepSeek-V3.1-Think to provide answers in a shorter time compared to its predecessor, DeepSeek-R1-0528 [3]. - Improved agent capabilities through post-training optimization have led to better performance in tool usage and intelligent tasks [3]. Group 2: API and Pricing Changes - The official app and web model have been upgraded to DeepSeek-V3.1, allowing users to switch between thinking and non-thinking modes via a "deep thinking" button [3]. - The DeepSeek API has also been upgraded, with deepseek-chat corresponding to non-thinking mode and deepseek-reasoner to thinking mode, expanding context to 128K [3]. - Starting from September 6, 2025, the pricing for API calls will be adjusted, with the cancellation of night-time discounts [4][6].

智通财经APP获悉，8月21日，DeepSeek正式发布 DeepSeek-V3.1。本次升级包含主要变化有：混合推理架构(一个模型同时支持思考模式与非思考模式)； 更高的思考效率(相比 DeepSeek-R1-0528，DeepSeek-V3.1-Think 能在更短时间内给出答案)；更强的 Agent 能力(通过 Post-Training 优化，新模型在工具使 用与智能体任务中的表现有较大提升)。 表 2：搜索智能体测评(测试结果调用商用搜索引擎 API+网页过滤+128K context window；R1-0528 使用内部 workflow 模式测试；HLE 测试同时使用 python 与 search 工具) | Benchmarks | DeepSeek-V3.1 | DeepSeek- | | --- | --- | --- | | | | R1-0528 | | Browsecomp | 30.0 | 8.9 | | Browsecomp_zh | 49.2 | 35.7 | | HLE | 29.8 | 24.8 | | xbench-DeepSearch | 71.2 | 55.0 | ...