RigRoute Labs · Independent research

Reviews

Explore RigRoute Labs hardware reviews and local AI research, with measured results, test conditions and limitations.

Hardware reviewPublished 28 September 2026Updated 30 September 2026

Two Radeon systems at 8K context depth: measured prefill and generation differences

RigRoute Labs measured prompt-processing and generation throughput for two models on an RX 7800 XT system and an RX 9070 system with llama.cpp Vulkan at 8,192 tokens of context depth. Evidence level E1; the host systems differ.

Read the article →

Local AI researchPublished 4 October 2026

Local AI on a 32 GB M1 Pro: fitting the model is only the start

We tested Qwen2.5 models from 3B to 32B on a 32 GB M1 Pro. Some long prompts took nearly eight minutes before the model began answering.

Read the article →