DeepSeek Releases V4.1-Flash-1017 AI Model Update With Faster Performance and Open Search APIs

DeepSeek Launches Enhanced AI Model With Performance Gains

DeepSeek released its updated AI model V4.1-Flash-1017 late Saturday, marking another rapid advancement from the Chinese AI startup as it challenges both domestic competitors and Silicon Valley giants. The Hangzhou-based company announced the update through a series of posts on X, highlighting stronger benchmark performance, faster task completion, and shorter outputs compared to its previous version1

1

. The model is now live across DeepSeek's web platform, mobile app, and developer API2

2

.

Benchmark Scores Show Measurable Improvements

The DeepSeek-V4.1-Flash-1017 demonstrates substantial improvements across reasoning, coding, agentic tasks, and visual benchmarks, according to the company's technical report published on Hugging Face1

1

. On Humanity's Last Exam, which measures AI performance across thousands of questions in math, sciences, and humanities, the revised model scored between 36.8 and 42.6, improving upon the previous V4.1-Flash score of 36.8 to 39.11

1

. The model achieved 77.0 on the DeepSWE v1.1 benchmark, up from 74.2, and reached 60.5 on Automation-Bench V1.0.6, compared with 54.8 previously2

2

.

Cost-Effective Performance Improvement Through Advanced Learning

DeepSeek described the update as a "Pareto improvement" in performance and cost, indicating gains in capabilities and efficiency without a corresponding increase in expense2

2

. The model leverages reinforcement learning with self-feedback to improve end-to-end project development, including building games and creating 3D models from scratch2

2

. The system has been optimized for tools including DeepSeek Harness, Tencent CodeBuddy, and WorkBuddy2

2

. Enhanced capabilities now span uses ranging from game creation to 3D modeling, representing a minor but meaningful update to the earlier V4.1-Flash1

1

.

Search and Fetch APIs Now Available to All Users

Alongside the AI model update, DeepSeek opened its Search and Fetch APIs to all users, powered by the company's in-house search engine2

2

. Each Search or Fetch API endpoint costs $0.75 per 1,000 calls, continuing the company's strategy of aggressively pricing its services2

2

. This pricing approach has previously impacted competitors, with the September release of the slimmed-down V4.1-Flash causing shares of rivals including MiniMax Group Inc., Z.ai, and Alibaba Group Holding Ltd. to plummet on the Hong Kong stock exchange1

1

.

Infrastructure Challenges Amid Rapid Growth

The rapid pace of new launches from DeepSeek underscores the intensifying contest between China's open-weight models and the proprietary models of US counterparts like Anthropic PBC and OpenAI, which remain top performers but with far higher costs1

1

. However, earlier this week, DeepSeek suffered an outage of about five hours, leaving its servers unreachable. Some experts attributed the glitch to a lack of sufficient computing resources1

1

. Service disruptions and slowdowns have increased in recent months as DeepSeek's aggressively priced AI models attract developers, putting pressure on its infrastructure1

1

. Watch how DeepSeek balances its rapid model releases with infrastructure scaling, and whether its pricing strategy forces established players to adjust their own cost structures in the coming months.