Natalie Miller

Perplexity AI has unveiled the R1 1776, a revamped version of the DeepSeek-R1 language model, now designed to offer unbiased, factual information without any censorship. This new release aims to fix some of the shortcomings of the original DeepSeek-R1, which often avoided sensitive topics or gave responses aligned with certain political agendas. While DeepSeek-R1 was already a powerful language model with strong reasoning skills, it sometimes avoided certain issues or provided responses that leaned toward Chinese Communist Party viewpoints. To address this, Perplexity AI developed R1 1776 to deliver more accurate and unbiased answers while maintaining its reasoning capabilities. The R1 1776 model has been retrained to remove any censorship, allowing it to tackle all kinds of questions directly and without bias. The Perplexity AI team compiled a diverse set of over 1,000 examples from multiple languages, including 40,000 real user questions about previously censored topics. This datas...

DeepSeek R1 is an advanced open-source AI model developed by the Chinese startup DeepSeek. It was launched in January 2025 and has quickly gained popularity as a credible rival to other AI models like ChatGPT. Key Features of DeepSeek R1: Open-Source: DeepSeek R1 is fully open-sourced under the MIT license, allowing researchers and developers to access, modify, and build upon its codebase. Advanced Capabilities: The model excels in various tasks, including mathematical problem-solving, coding, and natural language reasoning. Reinforcement Learning: DeepSeek R1 was developed using pure reinforcement learning, which allows the model to learn and improve without relying on supervised data. Cost-Effective: The development of DeepSeek R1 required less computing power and was achieved at a fraction of the cost compared to other models. Privacy: For users who prefer to run AI models locally to maintain privacy, DeepSeek R1 can be integrated with tools like Ollama, allowing users to host the m...