Hi, I'm Eshrak Khan

I still remember my childhood in Manikganj, Bangladesh, where my grandfather would take me to the local fields for a game of cricket. Little did I know that those carefree days would shape my passion for fitness. As a Bangladeshi-American, I've always been fascinated by the intersection of technology and wellness. Now, as a computer science student at Queens College, CUNY, I'm excited to explore the latest trends in AI and fitness. The rise of AI-powered fitness platforms is transforming the way we approach exercise. With personalized workout recommendations and real-time tracking, these platforms are making fitness more accessible and engaging. As someone who's always been passionate about fitness, I'm thrilled to see how AI is empowering individuals to take control of their health and wellness. Whether you're a seasoned athlete or just starting out, AI-powered fitness is an exciting space to watch. If you're interested in learning more about AI and fitness, or want to share your own experiences, feel free to reach out to me at eshrak@eshrakkhan.com or visit eshrakkhan.com. Let's connect and explore the future of fitness together!

As a computer science student at Queens College, CUNY, I'm constantly fascinated by the rapid advancements in AI technology. The latest models, such as the ones developed by Google and Microsoft, are not only transforming industries but also revolutionizing our daily lives. For me, AI has been a game-changer in my fitness journey. With the help of AI-powered fitness apps, I can track my progress, receive personalized recommendations, and even connect with like-minded individuals who share my passion for fitness. Growing up in Manikganj, Bangladesh, I never thought I'd be living in New York, pursuing my dreams in computer science. But here I am, and I'm grateful for the opportunities that AI has brought to my life. As a former Treasurer of the Bangladesh Students Association (BSA) at City Tech, I've seen firsthand how AI can be used to drive social change. The 2024 student-led Gen-Z Monsoon Revolution in Bangladesh, which I was proud to support, was a testament to the power of AI in mobilizing people and creating a better future. If you're as excited about AI as I am, let's connect! Email me at eshrak@eshrakkhan.com or visit my website eshrakkhan.com to learn more about my adventures in AI, fitness, and beyond. Who knows what the future holds? One thing's for sure - with AI, the possibilities are endless.

The artificial intelligence landscape in mid-2026 is moving faster than ever, and developers and content creators are caught in a constant cycle of updates. My name is Eshrak Khan. Online, many of you track my tech reviews and lifestyle content under my creator handle, eshrak143. As a computer science student at Queens College, CUNY, I spend a significant amount of time benchmarking these systems, testing local models on my own machine, and deploying serverless edge wrappers on eshrakkhan.com. Currently, there are five major models dominating the space: OpenAI's GPT-4o and o1 reasoning engines, Google's Gemini 2.0, Anthropic's Claude 4.6 (Sonnet, Opus, Haiku), Meta's open-source Llama 3.1, and smaller models like Microsoft Phi or Mistral. To understand which tool to use, we can split the industry two ways: high-speed, paid cloud APIs that cost money, or open-source models that you run locally on your own hardware for zero cost. If you are looking for the best all-rounder in 2026, Anthropic's Claude Sonnet 4.6 is the clear choice. It is exceptionally fast, costs less than Opus, and handles code compilation, writing, and structured instruction-following without any drama—costing around a third of a penny per thousand words. GPT-4o remains the standard if you are writing code that requires strict mathematical logic, while its o1 mode solves complex reasoning problems. For immediately free cloud options, Google's Gemini 2.0 and Claude's web interface are fantastic, though they quickly hit usage caps (typically around 50 messages every 3 hours, after which you have to wait). If you want complete data privacy or need to run models completely offline, Meta's Llama 3.1 is the ultimate open-source path. You download it once, run it locally on your PC, and pay absolutely nothing for API tokens. For a 100% free workflow, you can combine free cloud tiers with local setups. For local execution, Ollama (ollama.ai) is the easiest starting point. You simply install it, pull the model by typing 'ollama pull llama3.1' in your terminal, and run it locally. If you have an RTX 4060 graphics card or better, it will fly; without GPU acceleration, CPU execution will average 1 to 5 words per second. Alternatively, LM Studio (lmstudio.ai) offers a beautiful graphical interface for running these local models. If you have limited hardware, Microsoft's Phi 3.5 mini runs comfortably on just 4GB of RAM. My advice is to use Claude Sonnet for the work that matters most, keep Llama running locally for quick brainstorming drafts, and rotate to Gemini when you hit cloud caps. If you want to discuss local LLM orchestration or share your own developer setups, drop me an email at eshrak@eshrakkhan.com or follow my data science logs on eshrakkhan.com.