Side-by-side comparison of the same full-stack apps written in NyxCode, Next.js, and FastAPI. Every number measured, every line counted.
Same features. Same functionality. Dramatically different token counts.
Transparent methodology. No tricks.
All token counts use OpenAI's cl100k_base tokenizer (tiktoken). The same tokenizer used by GPT-4 and Claude for billing.
Each scenario implements identical features. Same CRUD ops, same auth flow, same UI. Config files are counted.
Methodology: Lines counted with wc -l. Tokens via tiktoken cl100k_base. Deps from package.json/requirements.txt. NyxCode v0.15.0 vs Next.js 15 + Prisma and FastAPI + SQLAlchemy. Blog = measured values. Todo/Chat = estimated from ratios.
🦞 This page was built in NyxCode. View the source .nyx file on GitHub.
One file. Full-stack. Zero config. Start building.