Small LLM on Intel i5: 36 Tokens/Second Reality
Testing five small models on an Intel i5 revealed the LFM2.5-350M hits 36 tokens/sec, bypassing memory limits for local CPU inference.
Sofia Berg
11 min read