Count a text's tokens and estimate the cost of one call on popular LLMs (or the models you list), input plus an expected number of output tokens, and whether it fits each context window.