micsapp Openrouter Pricing Basics
Overview
This skill explains OpenRouter's per-token pricing model, how to calculate costs for different models, and strategies for staying within budget.
Prerequisites
- OpenRouter account with credits
- Understanding of token concepts (prompt vs completion tokens)
Instructions
- Understand the pricing model: OpenRouter charges per token with separate rates for prompt (input) and completion (output) tokens, listed as price per 1M tokens
- Query model pricing: Use the
/modelsendpoint and inspectpricing.promptandpricing.completionfields for each model - Estimate request costs: Calculate cost as
(prompt_tokens * prompt_price + completion_tokens * completion_price)using theusageobject from each response - Track cumulative spend: Use the
X-OpenRouter-Costresponse header or query/api/v1/auth/keyto check remaining credits - Optimize costs: Use cheaper models for simple tasks (e.g.,
google/gemma-2-9b-it:free) and reserve expensive models for complex reasoning
Output
- Cost-per-request calculations for your chosen models
- Credit balance and usage tracking setup
- Model pricing comparison table sorted by cost
Error Handling
| Error | Cause | Fix |
|---|---|---|
| 402 Payment Required | Insufficient credits for the chosen model | Top up credits or switch to a cheaper/free model |
| Unexpected costs | Completion tokens exceeded max_tokens | Always set max_tokens to cap output length |
| Price mismatch | Using outdated pricing data | Re-query /models endpoint; pricing updates dynamically |
See ${CLAUDE_SKILL_DIR}/references/errors.md for full error reference.
Examples
See ${CLAUDE_SKILL_DIR}/references/examples.md for runnable code samples.