Use-case calculators
Start with the workload, then compare models.
Each guide uses a concrete token scenario, shows low-cost candidates from the checked-in model database, and links into compare presets for the next decision.
Short turns
Chatbot
Estimate support or product chat costs from short input, short output, and repeated monthly messages.
500 input tokens, 300 output tokens, 100 messages per user, 1,000 monthly users
Context-heavy answers
RAG
Check whether low-cost models can handle retrieved document context before you compare candidates.
100 question tokens, 2,000 retrieved tokens, 500 answer tokens, 10,000 monthly questions
Long inputs
Summarization
Compare input-heavy costs for reports, articles, notes, and other long-document workloads.
10,000 input tokens, 500 output tokens, 1,000 documents per month
Repeated repo work
Coding agent
Price longer prompts, more context, larger outputs, and multi-step coding-agent iterations.
2,000 prompt tokens, 5,000 context tokens, 1,500 output tokens, 3 iterations, 500 monthly tasks
Need a custom token count?
Use the calculator after you pick a starting scenario, then adjust input, output, volume, and cache assumptions.