Archive · 30 May 2025
DeepSeek R1 Model Update Claims to Rival ChatGPT, Raises Censorship Concerns
DeepSeek, a Chinese AI company, updated its R1 model, claiming it now performs similarly to top AI models like ChatGPT.
The updated model, named R1-0528, is better at solving problems, reasoning, and offers better coding experience.
However, a developer is worried that the new model is heavily censored when it comes to certain topics, especially criticism of the Chinese government, and has difficulty responding to sensitive subjects.
This has raised concerns about free speech and censorship.
DeepSeek has not yet responded to these censorship claims.
DeepSeek updated its R1 AI model with improvements.
The updated model is named R1-0528.
The company claims R1-0528 rivals ChatGPT O3 and Gemini 2.5 Pro.
The upgraded model shows improvements in reasoning, math, logic.
A developer alleges censorship in the new model regarding sensitive topics.
- Who
- DeepSeek, a Chinese AI startup, developed the R1 model.
- What
- DeepSeek updated its R1 AI model, claiming performance improvements.
- When
- The update to R1-0528 was recently released.
- Where
- The model was developed in China. The developer conducted a test on a website called SpeechMap.
- Why
- DeepSeek aimed to improve R1's reasoning, math, logic, and reduce hallucinations.
Developer's Concerns
DeepSeek's Claims
Performance & Censorship
Developer's Concerns
A developer alleges the new DeepSeek R1 model is heavily censored, particularly regarding criticism of the Chinese government.
DeepSeek's Claims
DeepSeek says the upgraded R1 model shows significant improvements in handling complex reasoning tasks, math, programming, and general logic.
- Developer
- China
- Upgraded Model Name
- R1-0528
- Model Improvements
- Improved reasoning, math, logic, less hallucination
- Performance Claim
- Approaching leading models, such as O3 and Gemini 2.5 Pro