Compare monthly spend across Claude, Gemini, GPT, Kimi, GLM and mini-tier models, then get a ranked answer to the question that actually matters: what should you change first, output tokens, retries, long context, premium-model overuse, or missing caching.
Your workload
Change this first
Estimated monthly spend on your current model
All levers, ranked by estimated monthly savings
Same workload priced across models
Model
In $/1M
Out $/1M
Cached in $/1M
Monthly cost
Scale
Sample list prices per million tokens, editable in spirit: treat them as reference points and re-check vendor pricing pages before committing. Costs assume 30 days and apply your retry multiplier and cache hit rate.