DeepSeek is recruiting for 17 positions, with R&D focus shifting toward Agent productization. Positions include algorithm research, data evaluation, and infrastructure engineers, focusing on reinforcement learning and evaluation system design. The company also values experience with AI coding tools. Compared to the deep learning researcher positions from the beginning of the year, this recruitment clearly tilts toward Agent productization.

DeepSeek opened 17 recruitment positions on March 25, with core R&D direction shifting toward Agent productization, including algorithm research, data evaluation, and infrastructure engineering. Recruitment requirements emphasize the use of AI coding tools and establish a dedicated product manager position for the Agent direction, demonstrating the company's emphasis on Agent technology. Compared to January, this recruitment clearly leans toward Agent productization.

Rakuten Group launched Rakuten AI 3.0, a Japanese-specialized large language model, on March 17, claiming to surpass GPT-4o on multiple Japanese language tests. However, netizens discovered that the model may be based on DeepSeek development and raised questions about its pro-China stance and the authenticity of its independently developed technology, sparking discussions.

Rakuten Group announced the release of Rakuten AI 3.0 on March 17, a high-performance AI model with 671 billion parameters optimized for Japanese. The company claims that it outperforms GPT-4o on multiple benchmark tests. The model is freely open-sourced under the Apache 2.0 license, fine-tuned based on the DeepSeek V3 model, and received training computational support from the Japanese government.

OpenRouter launched two anonymous models, Hunter Alpha and Healer Alpha, on March 12, both available for free. Hunter Alpha is a 1 trillion parameter text model focused on long-term planning and complex reasoning; Healer Alpha is a multimodal model supporting visual and audio inputs. The community has various speculations about the identities of these models, but no official confirmation has been provided.

Gate News Report, March 11 — According to the latest data from Polymarket, the market odds that DeepSeek V will be released on March 31 are 42%. Currently, the trading volume on this prediction market has exceeded $1.04 million.

On March 10th, a16z released the Top 100 Global AI Consumer Products list, showing that the AI market is divided into three major camps: the West, China, and Russia. Western products are mainly used in the United States and other regions, China is led by companies like Doubao, and Russia's Yandex Browser performs outstandingly. DeepSeek spans all three camps, demonstrating a trend toward globalization.

BlockBeats News, March 3 — Research firm Quest Mobile released the "Core Report on AI Application Layer Development by 2025," which shows that by December 2025, the top five AI-native apps in the market by monthly active users are Doubao, DeepSeek, Yuanbao, Ant Afo, and Alibaba Qianwen. Ant Group's general AI assistant Lingguang, released in November last year, also entered the top ten. (Jin10)

The new paper released by the DeepSeek team introduces an inference system called DualPath, which optimizes large model inference performance, achieving a 1.87x increase in offline throughput and a 1.96x increase in online service. The application of large models is driving the evolution of forward intelligent agent systems, promoting a major transformation in human-computer interaction.