Model Speed
Domain: Models & Providers
Motivation
How fast a model responds. Measured in tokens per second. Varies by provider and model.
Concrete Example
GPT-4o: 80 tokens/sec. Claude: 50 tokens/sec. DeepSeek: 120 tokens/sec. Faster models feel more responsive in conversation.
Analogy
Internet speed - 100Mbps vs 1000Mbps. Faster is better. Model speed determines how quickly Hermes responds.