DeepSeek\u2019s most recent update, R1-0528, may be called a \u201cminor trial upgrade,\u201d but it makes major leaps. With considerable advancements in reasoning, code generation, and dependability, R1-0528 places DeepSeek as an open-source contender to Gemini 2.5 Pro. It even approaches OpenAI\u2019s o3 and o4-mini in terms of code benchmarks. In this blog, we\u2019ll explore the features of this upgrade and provide hands-on comparisons between DeepSeek R1 and the new R1.1.<\/span><\/p>\n DeepSeek R1-0528, also called DeepSeek R1.1 is a next-generation, open-source large language model (LLM) released by DeepSeek. This new model represents a new generation of AI systems going head-to-head with world-leading proprietary models by the likes of OpenAI and Google.<\/span><\/p>\n DeepSeek R1-0528 is particularly focused on reasoning, code generation, and complex problem-solving abilities. This makes it a game-changing resource for researchers, developers, and businesses seeking to leverage powerful AI systems, without relying on a closed ecosystem and proprietary, subscription-based AI models.<\/span><\/p>\n Want to learn more about the DeepSeek R1 model? Click here<\/a>.<\/em><\/p>\n DeepSeek R1 came out earlier this year with a bag of innovative features. Now, with its latest update, DeepSeek R1-0508, the open-source model has become bigger and better.<\/span><\/p>\n Some of its upgraded features are:<\/span><\/p>\n The earlier version of DeepSeek R1 blew the world away with its performance. It gave tough competition to all the popular models at the time and proved that open-source models<\/a> can be at par with closed-source ones. Now, DeepSeek R1.1 also proves to be quite similar in its impact!<\/span><\/p>\n Let\u2019s look at the performance of DeepSeek R1.1 against the top models on the composite LLM performance score, which is aggregated from benchmarks<\/a> like MMLU, HumanEval, GSM8K, BBH, TruthfulQA, etc.<\/span><\/p>\n With a median of 69.45, DeepSeek R1 \u2013 0528 performs reliably across a wide range of tasks (e.g., reasoning, coding, math, etc.). It delivers near Claude-level median performance at a fraction of the cost, making it one of the best value-for-money models in this list. It outperforms Gemini 2.5 Pro and even Claude Sonnet 4 in consistency while costing 5x\u20137x less.<\/span><\/p>\n Now let\u2019s get to a real-world comparison of DeepSeek R1 and R1-0528 on reasoning, code generation, and reliability. We\u2019ll be testing out both models on 3 different tasks to see how they perform and find out if the new upgrade actually does better.<\/span><\/p>\n Both models will be tasked with building an HTML page that mimics the Instagram design. This will test their code generation quality, structural understanding of UI, and correctness of logic in frontend development.<\/span><\/p>\n Input Prompt:<\/b> \u201c<\/span>Create a responsive HTML and CSS layout that looks like Instagram\u2019s main feed page. It should include:<\/span><\/i><\/p>\n A fixed top navigation bar with the Instagram logo on the left, a search bar in the center, and navigation icons (home, messages, explore, notifications, profile) on the right.<\/span><\/i><\/p>\n A sidebar on the left for navigation with icons and labels similar to Instagram (Home, Search, Explore, Reels, Messages, Notifications, Create, Profile).<\/span><\/i><\/p>\n A main feed in the center showing several post cards. Each post should include:<\/span><\/i><\/p>\n A sidebar on the right with suggestions for \u201cPeople you may know.\u201d<\/span><\/i><\/p>\n \nWhat is DeepSeek R1-0528?<\/span><\/h2>\n
What\u2019s New in DeepSeek R1-0528?<\/span><\/h2>\n
![\"DeepSeek](\"https:\/\/cdn.analyticsvidhya.com\/wp-content\/uploads\/2025\/05\/GsEHdrIWcAA3qJw.webp\")
<\/figure>\n\n
DeepSeek R1-0528: Performance Benchmarks<\/span><\/h2>\n
![\"DeepSeek](\"https:\/\/cdn.analyticsvidhya.com\/wp-content\/uploads\/2025\/05\/final_llm_comparison_table.webp\")
DeepSeek R1-0528 vs DeepSeek R1: Hands-on Comparison<\/span><\/h2>\n
Task 1: Construct a User Interface Similar to Instagram<\/span><\/h3>\n
\n
Output by DeepSeek R1-0528:<\/span><\/h4>\n