I compared the standard API list prices of 18 models from OpenAI, Anthropic, Google, xAI, DeepSeek, and Mistral. To make the numbers easier to understand, I calculated the cost of the same sample workload across different models: 100,000 input tokens 20,000 output tokens Standard short-context pricing No batch, caching, tool-use, or priority-processing discounts Approximate cost for this workload: Gemini 2.5 Flash-Lite: $0.018 DeepSeek V4 Flash: $0.0196 Mistral Small ...