Search AI Products and News

A new study led by Apple researchers has poured cold water on the current large reasoning models (LRMs) that are highly anticipated. The research found that when solving complex tasks, reasoning models designed to simulate thinking processes, such as Claude3.7Thinking and Deepseek-R1, not only failed to show advantages but also exhibited serious problems like insufficient reasoning and performance collapse. This study tested four classic logical puzzles: the Tower of Hanoi, Checkers, River Crossing, and Block World. These puzzles allow precise control over the tasks.

Recently, ByteDance's Volcano Engine Large Model Service Platform, Volcano Ark, officially announced the integration of the latest version of DeepSeek-R1-0528. This move not only demonstrates ByteDance's technical strength in the field of large model services but also provides more efficient and convenient large model application experiences for enterprise users and developers. The Volcano Ark platform has built a high-performance service system targeting the core requirements of large model applications, namely speed and stability. By using its self-developed xLLM high-performance inference framework, the platform achieves a model latency as low as 30ms per T.

DeepSeek recently made significant updates to its high-performance reasoning AI model, DeepSeek-R1, greatly enhancing the model's performance in code generation and complex reasoning tasks, drawing widespread attention in the artificial intelligence community. Below is a comprehensive analysis of the key highlights of this update based on public information and the latest developments. R1 Model Update: Significant Improvement in Code Ability The latest update to DeepSeek-R1 has achieved remarkable breakthroughs in code generation capabilities. Testing shows that the new version of R1 significantly improves its ability to handle complex coding tasks.

DeepSeek officially announced that its R1 model has completed minor version trial upgrades and is now open for user testing. This upgraded model is named "DeepSeek-R1-0528", showing amazing performance in various tests. The new DeepSeek-R1-0528 has significantly improved programming capabilities, aesthetic design, and code completion. During testing, with just simple instructions, the model can quickly generate high-quality code and design works. In addition, R1-0528 has excelled in generating complex pre-

At the "FUTU Investment Expo" held on March 21st at the Hong Kong Convention and Exhibition Centre in Wan Chai, FUTU Securities officially announced the integration of DeepSeek-R1 and the launch of a series of upgraded artificial intelligence (AI) product features. The expo, themed "Intelligent" Investing, attracted significant attention from retail investors. FUTU Securities founder and chairman, Li Hua, stated at the expo that the newly launched Bull AI will provide investors with more flexible services. Unlike the previous unidirectional "AI stock interpretation", Bull AI is capable of processing...

SiliconCloud officially announced that Trae, ByteDance's first AI-native Integrated Development Environment (IDE), has integrated with the SiliconCloud platform to provide developers with a more efficient programming experience. Trae now integrates multiple SiliconCloud models; users can access various coding models such as DeepSeek-R1, V3, QWQ-32B, and Qwen2.5-Coder simply by using an API key. This flexible selection caters to diverse development needs.

Silicon-Based Flow's official WeChat account announced that, effective immediately, the DeepSeek-R1 & V3 APIs on the SiliconCloud platform now support batch inference (BatchInference). Users can send requests to SiliconCloud via the batch API, unaffected by real-time inference rate limits, with task completion expected within 24 hours. Compared to real-time inference, DeepSeek-V3 batch inference pricing is reduced by 50%. From March 11th to March 18th, DeepSeek-R1 will enjoy a 75% price reduction.