DeepSeek has begun offering an updated artificial intelligence model that the Chinese startup says beats American AI tools on some measures, even though it costs far less to run. The Hangzhou-based company released its V4.1-Flash-1017 update late Saturday afternoon, Oct. 17. Nikkei Asia reported that the new version is said to outperform models from OpenAI and Anthropic by some measures. For a startup that built its name on low prices, that’s a bold public claim.
The company announced the update in posts on X, saying the model offers “stronger benchmark performance, faster task completion, and shorter outputs than the previous version.” It’s available on the web, through the DeepSeek app and on its developer platform. Developers watching their budgets can test it without signing a contract with a large cloud vendor. That’s part of the appeal.
The Benchmarks
DeepSeek put numbers behind its claims. On Humanity’s Last Exam, a test covering thousands of questions in math, sciences and humanities, the revised model scored between 36.8 and 42.6. The earlier V4.1-Flash scored between 36.8 and 39.1. The gain is modest. The company described the release as a minor update, meant to improve uses ranging from game creation to 3D modelling. Still, DeepSeek said the latest iteration made “substantial improvements across reasoning, coding, agentic tasks, and visual benchmarks,” according to a technical report posted on Hugging Face.
The update came only weeks after DeepSeek unveiled the slimmed-down V4.1-Flash system in early September. The company offered that model at a steep discount, and markets noticed. Shares of MiniMax Group Inc, Z.ai and Alibaba Group Holding Ltd, which drives the Qwen models, plummeted on the Hong Kong stock exchange. Investors who’d priced in a slower pace of Chinese competition got a reminder: a cheaper product can reshape the field quickly.
Competitive Pressure
The update extends DeepSeek’s challenge to Silicon Valley rivals Anthropic PBC and OpenAI, as well as domestic competitors Moonshot Kimi and Z.ai. The rapid release schedule reflects a contest between China’s open-weight models and proprietary U.S. models. American systems remain top performers, but they cost far more. That gap sits at the heart of the business case. If a model delivers comparable results at a fraction of the price, customers have every reason to switch. No amount of marketing can fully protect a premium price tag.
DeepSeek’s pricing strategy has created problems, too. The company suffered an outage of about five hours that left its servers unreachable. Some experts attributed the glitch to insufficient computing resources. Developers flocked to the company’s aggressively priced models, and slowdowns and service disruptions grew more frequent in recent months. A low price draws customers quickly, but it can stretch a company thin when demand outruns its capacity to serve them.
Security and Market Implications
The episode shows how quickly the market can move when a Chinese firm undercuts American pricing. For U.S. AI companies, the threat is commercial for now, but it also raises questions about where the next generation of cheap, capable AI tools will be built and who will control the infrastructure beneath them. Free-market competition is doing what it does best: forcing rivals to justify their costs. Whether the outcome serves American interests depends on whether domestic firms can keep innovating faster than lower-priced challengers.
Why This Matters: The pricing pressure is a reminder that competition, not government mandate, is setting the terms in artificial intelligence right now. DeepSeek’s ability to undercut rivals while claiming performance gains puts a spotlight on American companies’ cost structures and the investors who back them. The stock slide in Hong Kong shows that the consequences reach beyond a single product launch. The five-hour outage and the capacity strain behind it also reveal the operational limits of aggressive pricing, and how a company’s growth can outpace its infrastructure. For policymakers and business leaders alike, the central question is whether American firms can sustain their lead on quality and price in a market where a cheaper rival can ship an update in days. The answer will shape investment decisions, the profitability of AI providers and the broader competitive position of the United States in a technology rapidly becoming central to commerce and national strength.
- deepseek
- v4.1-flash-1017
- artificial intelligence
- ai models
- openai
- anthropic
- china
- ai benchmarks