The latest articles on DEV Community by harish venkatakrishnan (@harish_venkatakrishnan_2c). https://dev.to/harish_venkatakrishnan_2c https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F3868320%2Ff7ee4b56-9331-4ccb-bc79-9c16a78ec060.png https://dev.to/harish_venkatakrishnan_2c en harish venkatakrishnan Wed, 08 Apr 2026 17:36:48 +0000 https://dev.to/harish_venkatakrishnan_2c/i-benchmarked-gemma-4-e4b-against-the-gemma-family-here-are-the-results-gpo https://dev.to/harish_venkatakrishnan_2c/i-benchmarked-gemma-4-e4b-against-the-gemma-family-here-are-the-results-gpo <p>I've been running real benchmarks on open-source LLMs to test things the big labs don't publish. My latest experiment compares Gemma 4 E4B against the rest of the Gemma family on enterprise tasks.</p> <p>Full results with methodology and limitations: <a href="https://aiexplorer-blog.vercel.app/post/gemma-4-e4b-enterprise-benchmark" rel="noopener noreferrer">https://aiexplorer-blog.vercel.app/post/gemma-4-e4b-enterprise-benchmark</a></p> <p>I'm building a collection of these experiments at aiexplorer-blog.vercel.app/experiments — covering structured JSON output, context position bias, RAG compliance, and prompt injection defenses.</p> <p>Feedback welcome — what models or tasks should I benchmark next?</p> ai machinelearning opensource showdev