Chat with KeyLM here to test it out!
A 5M hybrid-looped LLM with effort levels
Benchmark base LMs (10M–1B) on CPU — speed and quality
Explore tiny language model benchmarks
Compact model arena with GLM and GPT OSS commentary