Data guides
A small editorial layer over the live ModelRigs catalogue. Every performance claim keeps its measured or estimated label.
Guides
Best hardware for local LLMs by budget
Compare new local AI hardware in distinct US price bands, with model fit and clearly labeled speed estimates.
Best 128 GB machines for local AI planned
Compare 128 GB-class systems by fit, speed and cost.
Mac vs NVIDIA for local LLM inference planned
A workload-specific comparison of memory capacity, runtime support and throughput.
RTX 3090 vs 4090 for local LLMs planned
Measured and estimated local-inference evidence for two widely used GPUs.
How much VRAM or unified memory do local LLMs need? planned
Model size, quantization, KV cache and practical headroom explained with live fit data.