#DeepSeek
318 related articles
DeepSeek V4.1 Flash 1017 Hands-On: The Strongest Open-Source Model or a Distillation Mystery?
DeepSeek V4.1 Flash 1017 tested: Terminal Bench beats GPT 5.6 Soul, 3D and front-end generation near flagship, priced at 1/26 of Opus. Benchmarks, cost-performance, distillation doubts, and search API toolchain analyzed.
30 Yuan, 5 Hours: Building a Narrative Card Survival Game with DeepSeek-V4.1
A Bilibili creator used the DeepSeek-V4.1-Flash-1017 model to build the narrative card survival game Don't panic! in just 30 yuan and 5 hours, featuring 7 doomsday scenarios and humorous trivia, exploring new possibilities for AI game development.
DeepSeek V4.1 Flash Hands-On: 3D Modeling Impresses, But Too Early to Challenge GPT
DeepSeek V4.1 Flash (1017) hands-on: huge 3D modeling gains, beating closed-source models in some cases, but weak web aesthetics and token use up to 10x GPT's.
DeepSeek V4.1 Flash Hands-On: 3D Capabilities Transformed, Full Test Suite Cost Only $4
DeepSeek V4.1 Flash 1017 hands-on: 3D generation far surpasses the previous version, with Corvette modeling and C++ basketball games standing out, all tests costing just $3.93.
US Federal Agencies Ban DeepSeek: National Security Concerns Escalate
The US orders federal civilian agencies to remove DeepSeek products and third-party services built on its technology, citing national security concerns. This article analyzes the scope of the ban, its potential impact on DeepSeek, and new challenges for AI decoupling and open-source models.
DeepSeek Open-Sources Its Inference Engine: Lowering the Barrier to LLM Deployment
DeepSeek open-sources its inference engine, letting developers reproduce the official inference path locally. We analyze the real value, community reaction, and practical advice.
DeepSeek Open-Sources DeepInfer: A High-Performance LLM Inference Framework
DeepSeek open-sources DeepInfer, a high-performance, large-scale inference framework. This article analyzes its impact on private inference, community engines like gufo/halogen, and the open-source inference ecosystem.
DeepSeek V4.1 Flash 1017 Upgrade Analysis: Seamless Model Replacement and Cost-Reduction Practices
DeepSeek V4.1 Flash 1017 upgrade analysis: benchmark gains, seamless model ID replacement, LU Cloud price cuts, and how to migrate without breaking old configs.
DeepSeek V4.1 Flash 1017 Hands-On: Token Efficiency Fixes Its Biggest Weakness
DeepSeek V4.1 Flash 1017 tested: Terminal Bench rises from 31 to 42, reasoning tokens cut by about one-third, and coding task cost drops from 50 cents to 20 cents at the same unit price.
DeepSeek V4.1-Flash-1017 Silent Upgrade: Three Pitfalls Developers Must Know
DeepSeek silently swapped deepseek-flash for V4.1-Flash-1017 on Oct 17. Three pitfalls: thinking-mode params silently ignored, missing reasoning_content triggers HTTP 400, and Search/Fetch billing from Nov 1. Includes peak-pricing conversion for North America and per-turn cost estimates.
DeepSeek API Pricing Explained: Cost per Million Tokens Across 6 Models
DeepSeek API bills by input, cached input, and output tokens. See per-million-token prices for all 6 models, official vs. channel quotes, real cost examples, and caching savings.
DeepSeek V4.1 Flash 1017 (Max) Ignites Hacker News: Re-examining Model Version Naming and Iteration Logic
DeepSeek V4.1 Flash 1017 (Max) trends on Hacker News, sparking debate over model version naming, Flash/Max suffix meanings, and version iteration usability. This article reviews community focus, analyzes the real impact of naming confusion on developers, and highlights current information gaps.
Building a Web Research Tool with the DeepSeek Search and Fetch APIs
DeepSeek opens its Search and Fetch APIs at $0.75 per 1,000 successful requests. Build an inspectable web research tool in pure Python, covering search, fetch, source preservation, cost control, and prompt injection defense.
DeepSeek V4.1 Flash Migration Guide: Avoiding 400 Errors and Silent Model Swaps
DeepSeek V4.1 Flash 1017 migration in practice: fix the reasoning_content 400 after tool calls, detect the silent deepseek-flash swap, handle double peak billing, and connect Claude Code, Codex, and DeepSeek Harness.
DeepSeek V4.1 Flash 1017 Tested: Intelligence Rises to 45, Cost per Task Drops 47%
Artificial Analysis's independent evaluation shows DeepSeek V4.1 Flash 1017's intelligence index rising from 39 to 45 and cost per task falling from $0.49 to $0.26. The drop comes from fewer output tokens, not lower unit prices; actual spend also depends on peak pricing and the reasoning_effort level.
DeepSeek Cuts Prices on Two Fronts: Search API as Low as $0.75
DeepSeek releases V4.1 Flash 1017 and a Search & Fetch API at $0.75 per 1K requests—6.7× cheaper than Perplexity—plus why prediction markets are lagging.
DeepSeek-V4.1-Flash Lands on Ollama: Stronger Agentic Performance at Lower Cost
DeepSeek-V4.1-Flash-1017 is now on Ollama and Ollama Cloud, emphasizing strong agentic performance and lower cost per task, with US-based hosting and zero data retention. This article breaks down integration, cost framing, and compliance selling points, while noting that model parameters, context length, and other key details still await official confirmation.
DeepSeek Silently Swapped Models: The KV Cache Arithmetic Is the Real Announcement
DeepSeek silently swapped deepseek-flash for the V4.1-Flash-1017 checkpoint, calling it a minor update while benchmarks rose across the board. Starting from the public config.json, this article recomputes KV cache cost per token, explains why the 890-byte figure is a category error, and unpacks the pricing logic behind the 50x cache discount and the risks of silent model-alias swaps.
Build an AI Video-to-Quiz Tool in 3 Days: Next.js 15 + DeepSeek in Practice
A developer built an AI video-to-quiz tool in 3 days: Next.js 15 + DeepSeek-V3 turns YouTube lectures into 5 timestamped quiz questions with Anki export. Breaks down subtitle scraping fallbacks, prompt anti-hallucination, and active recall productization logic.
DeepSeek V4 Pricing Revealed: Flash Variant Ultra-Low Cost, New Gemini Model Surfaces
This week's LLM pricing watch: DeepSeek V4 Flash input just $0.30/M, Pro at $1.32/M; Gemini 3.8 Flash output hits $4.125/M, above expectations; four new model IDs including Gemini 4 Argon surface, plus selection advice.