AI API pricing
AI API Pricing Calculator
Compare estimated monthly API cost across major LLM models from the same request volume, input tokens, output tokens, and cached-input assumptions.
Compare AI API pricing
Choose a workload, enter monthly requests and token size, then compare provider cost before committing.
- 1Pick workloadUse the closest API usage scenario.
- 2Enter tokensAdd request volume, input, output, and cache.
- 3Compare modelsCheck monthly cost and effective rate.
- 4Share or exportCopy the link or download the CSV.
Next: copy the share link or export the CSV, then request a worksheet or send a pricing correction if the assumptions need review.
| Provider | Model | Input / 1M | Cached / 1M | Output / 1M | Discount | Monthly cost | Cost / request |
|---|
Same workload, many models
Keep request volume and token assumptions constant while comparing provider-level cost.
Account for repeated context
Apply a cached-input share when long instructions or retrieval context repeat across requests.
Drill into details
Use dedicated OpenAI, Claude, or Gemini pages when provider-specific billing rules matter.
When to use this AI API pricing calculator
Use this comparison when you are choosing a model family or estimating a new feature. It works best after you have a rough average for input tokens, output tokens, and monthly request volume.
For provider-specific rules like Claude prompt cache writes, Gemini grounding, Mistral batch discounts, xAI tool pricing, or OpenAI tokenization, use the dedicated calculator pages and source links after this first comparison.
| Need | Best page | Why |
|---|---|---|
| OpenAI-specific budget | OpenAI API Cost Calculator | Includes OpenAI cached input and budget buffer assumptions. |
| Claude prompt caching | Claude Token Cost Calculator | Separates cache reads, cache writes, and output cost. |
| Gemini grounding or Batch API | Gemini Cost Calculator | Includes context caching, Batch API, storage, and grounding assumptions. |
| Raw text token estimate | Token Counter | Starts from pasted prompt text before you know average token counts. |
FAQ
Is the cheapest model always the best choice?
No. Cost is only one input. Evaluate latency, context window, quality, tool support, data controls, and reliability before choosing a production model.
Why do cached input prices matter?
Repeated system prompts, policies, examples, or retrieval context can materially change cost when a provider supports discounted cached input.
Sources
Prices in this site were last checked on 2026-07-04. Pricing can change without notice.