GPT-4o vs Claude 3.5 Sonnet vs Gemini 1.5 Pro: real API cost comparison for production LLM apps (opens in new tab)

GPT-4o is the middle ground in this comparison: cheaper than Claude 3.5 Sonnet, more expensive than Gemini 1.5 Pro on short prompts, and still current for production use. Claude 3.5 Sonnet has the highest output-token cost here, which matters a lot for chatbots, coding agents, and any workload that generates long answers. Gemini 1.5 Pro looked cheapest on paper for prompts up to 128K tokens, but its price doubled above that threshold, and it was primarily attractive when you needed very large...