Complementary readings
zigllm is not the only educational from-scratch LLM project, and these comparisons are not adversarial. Each of the projects below is a different vantage point on the same object. Read them next to zigllm.
- vs. llm.c
zigllm vs. llm.c
Karpathy's llm.c is the C-from-scratch counterpart to nanoGPT — pure C, no dependencies, training included. zigllm is its Zig cousin, inference-only, with more architecture families and SIMD-first kernels.
Read llm.c if you want to see training-from-scratch in raw C. Read zigllm if you want to see inference-from-scratch in Zig with manual memory and quantization.
Read comparison → - vs. nanoGPT
zigllm vs. nanoGPT
Karpathy's nanoGPT is the canonical pedagogical GPT in Python. zigllm is the same impulse — but in Zig, with manual memory, and 18 architecture families instead of one. They are complementary readings, not competitors.
Read nanoGPT first to see the math without distractions. Read zigllm next to see what the math looks like once you take the runtime away.
Read comparison →