I built a coding-LLM comparison from 11 benchmark boards—looking for feedback

Choose a coding LLM by repository engineering, code generation, agentic evidence, open weights, context window, and API cost—not by one raw benchmark…