DeepSeek R1 0528 Update: New Features, Performance Gains Over OpenAI o3
DeepSeek quietly launched the R1 0528 model, which early testers report matches OpenAI’s o3 in benchmarks and style, while adding deeper chain‑of‑thought reasoning, better writing output, and extended thinking windows, and the announcement is followed by a promotion for the GOPS Global Ops Conference.
On May 29, DeepSeek released the latest R1 0528 version without any official announcement, simply opening the model to the public, with a detailed model card expected soon.
Early testers on the Live CodeBench platform found its performance comparable to OpenAI’s latest o3 model, and its output style similarly close to OpenAI’s o3.
Previously, DeepSeek announced in its official group that the R1 model had undergone a minor version upgrade, inviting users to test via the official website, app, or mini‑program, while keeping the API unchanged. A similar minor upgrade for the V3 model was announced in March.
According to reports from China Fund News and user feedback, the upgraded model shows a noticeable change in chain‑of‑thought (CoT) behavior.
Users highlighted several improvements: deeper reasoning akin to Google’s models, enhanced writing tasks with more natural language and better formatting, a unique fast yet thoughtful inference style, and extended thinking periods of up to 30‑60 minutes per task.
🌟0627【GOPS Global Ops Conference·Beijing】
AI Observability Platform Drives Intelligent Operations Upgrade – Large Models Lead Monitoring Data Innovation.
Recent recommended reads include articles on overtime in ops, data governance for AIOps, and ByteDance’s internal ban on AI coding tools.
Efficient Ops
This public account is maintained by Xiaotianguo and friends, regularly publishing widely-read original technical articles. We focus on operations transformation and accompany you throughout your operations career, growing together happily.
How this landed with the community
Was this worth your time?
0 Comments
Thoughtful readers leave field notes, pushback, and hard-won operational detail here.