AWS bill line item
USE1-Claude4.6Sonnet-input-tokens
Every token sent into a Bedrock model as part of the prompt (system instructions, chat history, retrieved context, few-shot examples) is metered and billed separately per model, distinct from output tokens.
Bedrock input tokens (per model)
Verified against official AWS documentation last checked
Why it shows up
Input-token spend climbs fastest when prompts carry more context than the task needs: long system prompts, unpruned chat history, or retrieval that stuffs in whole documents instead of relevant chunks.
Because the usage type is per-model, a silent model switch or an A/B test can double the number of these line items without the aggregate number appearing to move.
How to cut it
- Trim system prompts and cap retrieved context to what's relevant.
- Truncate or summarize conversation history instead of resending the full transcript each turn.
- Route repeated static context through prompt caching so it bills as a cache read, not full input.
- Use a smaller, cheaper model for steps that don't need frontier-level reasoning.
How to read this code
USE1 / USE2us-east-1 (N. Virginia) / us-east-2 (Ohio)
Source
Source: AWS docs →Found USE1-Claude4.6Sonnet-input-tokens on your own bill?
Drop your Cost Explorer CSV into the AWS Bill Analyzer. It'll flag line items like this one on your own bill and link straight back to the fix.
Analyze your own bill →Token spend moves every day.
Model prices and rate limits change between invoices. CostMon tracks your AI spend daily, so a change shows up before the bill does.