#DeepSeek
321 related articles
DeepSeek API Outage Lasts 4 Hours 52 Minutes: The Disaster Recovery Lesson Behind a Week's Consumption Compensation
DeepSeek API suffered a partial outage of 4 hours 52 minutes on October 15. The company fully refunded 7-day consumption as bonus credits. This article outlines the outage scope, compensation details, and provides a complete code solution for configuring a backup provider for DeepSeek calls, plus an action list for this week.
vLLM Releases v0.32.0rc2: DeepSeek-V4.1-Flash Evaluation Pinned to FP8-Engram Revision
vLLM releases v0.32.0rc2, with a CI change pinning DeepSeek-V4.1-Flash-NVFP4 evaluation to the FP8-Engram revision, reflecting the inference framework's focus on reproducible multi-precision evaluation.
DeepSeek Funding Doubles to 100 Billion Yuan, ByteDance Seed Establishes LLM Safety Department
DeepSeek's second funding round doubles to 100 billion yuan; ByteDance Seed becomes China's first to establish an LLM safety and alignment department. Huawei's Kirin 2027 chip enters testing, Tesla Optimus targets year-end mass production, and Yin Zhiyao warns of AI bubble and semiconductor industry consolidation.
AI Reasoning Breakthrough: The Logical Leap of DeepSeek-R1 and OpenAI
LLMs are shifting from probabilistic prediction to structured reasoning. This article analyzes OpenAI's chain-of-thought methods and DeepSeek-R1's reinforcement learning mechanism.
DeepSeek V4 Beginner's Guide: How to Choose Between Flash and Pro
What is DeepSeek V4? This guide covers web, App, and API usage, plus criteria for choosing between Flash and Pro, FAQs, and prompting tips for beginners.
DeepSeek Eyes Embodied AI: Liang Wenfeng in Frequent Talks with Founders
DeepSeek founder Liang Wenfeng is in frequent talks with embodied AI founders, planning to lead an investment in robotics startup Winverse AI. Meanwhile, Tencent Futian Lab welcomes former UBTech CTO Liao Hongtao, and Unitree's Wang Xingxing says the embodied AI critical point is near.
What Is DeepSeek Harness? A Complete Guide to LLM Harness Engineering
DeepSeek Harness developer preview released with an 'everything is a plugin' philosophy. This article explains the essence of Harness—the engineering system beyond the model, covering memory, tool calling, orchestration, and more.
dsh-desktop Hands-On: Is the Open-Source DeepSeek Harness Desktop Worth Trying?
dsh-desktop is an open-source desktop solution for the DeepSeek Harness plugin ecosystem with 30,574 stars, supporting self-hosting with no API costs. This article covers its core design, three onboarding paths, comparisons with similar tools, and hands-on experience with MonkeyCode.
DeepSeek V4.1 Flash: How Read-Write Separation Reshapes Long-Conversation Costs
DeepSeek V4.1 Flash uses read-write separation—shorter input paths, shared attention memory, and selective reconstruction—to restructure inference costs in long conversations.
DeepSeek + Spring AI in Practice: Building an Offline AI Family Doctor App from Scratch
A practical tutorial on building an offline AI family doctor app with DeepSeek + Spring AI: covers Ollama local deployment, Spring Boot 3 integration, role customization, and conversation persistence.
DeepSeek Harness Step-by-Step Tutorial: From Installation to Real-World Software Testing
A comprehensive guide to DeepSeek Harness: Node.js setup, NPX one-command install, DeepSeek & third-party model config, workspaces, web search, file I/O, and agent observability for software testing.
DeepSeek V4.1 Flash Real-World Test: Price Down 60%, So Why Did Total Cost Go Up 36%?
DeepSeek V4.1 Flash launches with 60% cache-hit price cut, but a 300M-token community test shows total cost rising 36%. We break down the asymmetric MoE architecture, benchmark gaps, and the silent routing controversy.
DeepSeek Harness Step-by-Step Installation Guide: From Environment Setup to First Run
Step-by-step guide to installing DeepSeek Harness: Node.js setup, CLI installation, API Key setup, workspace configuration, and desktop shortcut creation for beginners.
Testing DeepSeek Desktop Agent: Auto-Generate a Full Video for Just $0.35
A blogger tested DeepSeek Harness desktop Agent: auto-generated a full video for $0.35, built a daily briefing, and an app from plain English — all for $0.59 total.
Claude Code for Beginners: From Environment Setup to Connecting with Chinese AI Models
Complete beginner's guide to Claude Code: set up Node.js and Git Bash, install via npm, and connect DeepSeek to bypass China IP blocks. Includes HTML to-do app demo.
The Big Fat Fish Paradox: Why DeepSeek's Unofficial Mascot Must Never Be Enshrined
DeepSeek's unofficial mascot 'Big Fat Fish' sparked an orthodoxy war. This piece explores why this AI-user co-created open-source symbol is most at risk the moment it wins.
DeepSeek 4.1 Flash Released: A High-Speed Multimodal Reasoning Model
DeepSeek 4.1 Flash is a new multimodal open-source model supporting image understanding, text, and coding. Learn about its architecture and how to try it via serverless inference.
DeepSeek Open-Sources Six Core Components: Laying the Foundation for Domestic AI Chips
DeepSeek open-sources six production-grade infrastructure components for Huawei's Ascend platform — TileLang, DeepGEMM, DeepEP, FlashMLA, and more — closing the software ecosystem gap for domestic AI chips.
HarmonyOS Hits 90 Million Devices: Mate 90 and Ascend Ecosystem Breakthroughs
HarmonyOS hits 90M devices, DeepSeek open-sources full Ascend toolkit vs CUDA, and Mate 90 launches with the Kirin 9050 Pro logic-folded architecture chip.
DeepSeek Harness Full-Multimodal Desktop Setup Guide: Image Generation, Video, and Remote Control All in One
Step-by-step guide to building a full-multimodal DeepSeek Harness desktop: DSH Desktop client, Agnes free image/video generation, DeepSeek V4.1 Flash vision, AnySearch web search, and Anywhere remote mobile control.