Kimi K2.6 突击发布：月之暗面截击 DeepSeek V4，万亿 MoE 编程战升级
月之暗面突然发布 Kimi K2.6，主打万亿参数 MoE 与极致编程能力，剑指即将发布的 DeepSeek V4。两家中国 AI 巨头的万亿级模型之战正式打响。
GitHub Dual Update! DeepSeek Open-Sources Tile Kernels & DeepEP V2, Spoilers for V4's Infrastructure
DeepSeek has officially open-sourced Tile Kernels and updated DeepEP to V2. These low-level libraries not only push GPU efficiency to its limits but also reveal key architectural details of V4, including Engram and Manifold HyperConnection (mHC). The future of domestic AI is being rebuilt from the operator level.
DeepSeek-V4 Preview Officially Released: The Era of 1M Context and Dual-Model SOTA is Here
DeepSeek launches V4-Pro and V4-Flash preview versions, featuring a 1.6T MoE architecture, 1M context support, and leading performance in coding and reasoning.
DeepSeek Valuation Hits $20 Billion! Tencent and Alibaba in Investment Talks?
According to The Information, tech giants Tencent and Alibaba are in talks to invest in DeepSeek. The valuation of this domestic AI dark horse has reportedly soared to $20 billion, surpassing Moonshot AI (Kimi). What signals does this send as V4 approaches?
DeepSeek V4 Benchmark Leak! Allegedly Outperforming GPT-5 and Claude 4.7—Is Domestic AI Reaching the Global Peak?
Heavyweight news rocks the AI community: Alleged official DeepSeek V4 benchmark charts have leaked. Data suggests V4 has achieved a comprehensive lead over top-tier overseas models in reasoning and coding. While unconfirmed, the excitement is global.
Comprehensive Overtake! Alleged DeepSeek V4 Tech Report Leaks: LiveCodeBench 72.1, Domestic AI Dominance Era Begins
A screenshot from an internal DeepSeek technical report has rocked the globe. Data shows the upcoming V4 version comprehensively outperforming GPT-5.3 and Claude Opus 4.6 in coding, reasoning, and Agent orchestration. Here is a deep dive into the leaked benchmarks.
DeepSeek Valuation Hits $10 Billion! First Talks for External Capital—Nuclear Fuel for the V4 Era?
Cailianshe Scoop: DeepSeek is in its first-ever talks for external capital, with valuation soaring past $10 billion. This marks DeepSeek's evolution from a 'High-Flyer Quant Lab' to a global compute titan, building a capital moat for the upcoming V4 model.
DeepSeek Expands to Inner Mongolia! Self-Built Compute Factory Exposed—What Does This Massive Move Hint at on the Eve of V4?
DeepSeek has launched a large-scale physical infrastructure recruitment drive in Ulanqab, Inner Mongolia, with salaries up to 30,000 RMB. This signals a transition from renting cloud services to building its own data centers, constructing a computing moat for the massive inference and training needs of DeepSeek V4.
DeepSeek V4 slated for late April: Trillion parameters, million context length, and deep optimization with Huawei Ascend.
Latest information reveals that DeepSeek V4 will be officially released at the end of April 2026. This model will employ a 1 trillion parameter scale, support a 1M Token ultra-long context window, and has undergone deep low-level optimization specifically for Huawei Ascend 910B hardware.
DeepSeek V4 Gray Scale Testing Begins: 1 Million Long Context and Knowledge Base Updates Are Coming Soon
Multiple sources from X and Reddit confirm that DeepSeek has begun limited grayscale testing of the V4 model, supporting 1M context. The official release version is expected in late April.
DeepSeek V4: 1 Trillion Parameters, Multimodal, and the OpenClaw Agent Shift
Liang Wenfeng internally confirmed: DeepSeek V4 will be released in mid-to-late April. LTM breakthroughs may end the RAG era!
DeepSeek V4 Canary Leak: Fast Mode vs Expert Mode and the 2.6M Context Revelation
A deep dive into the leaked JSON configuration and early tests of DeepSeek's new "Fast" and "Expert" modes.
The Monster Opens Its Eyes! DeepSeek V4 'Vision Mode' Canary Leak: Native Multimodality vs Gemini 3.1
Deep diving into the leaked details of DeepSeek V4 Vision Mode, exploring how native multimodal architecture is reshaping AI spatial perception and agentic development.
GPT-6 Leak: April 14th Release! 40% Performance Boost, How Will DeepSeek V4 Respond?
DeepSeek V4 Release Date & Official Reveal: Why They Refused NVIDIA Early Access