Claude Sonnet 4.6 vs Claude Sonnet 5.5: price comparison
Claude Sonnet 4.6 costs $3.00 per million input tokens and $15.00 per million output tokens; Claude Sonnet 5.5 costs $2.00 and $10.00 (Anthropic list prices, checked 2026-11-12). This page compares cost only; it does not rate quality, so test both on your own task.
Prices and limits
| Claude Sonnet 4.6 | Claude Sonnet 5.5 | |
|---|---|---|
| Input (per 1M tokens) | $3.00 | $2.00 |
| Output (per 1M tokens) | $15.00 | $10.00 |
| Cache hit (per 1M) | $0.30 (10% of input) | $0.10 (5% of input) |
| 5-minute cache write | $3.75 | $2.50 |
| 1-hour cache write | $6.00 | $4.00 |
| Batch input / output | $1.50 / $7.50 | $1.00 / $5.00 |
| Context window | 1M | 1M |
| Max output | 128K | 128K |
| Minimum cacheable prompt | 1,024 tokens | 512 tokens |
| Tokenizer | earlier | newer (Opus 4.7 and later) |
Per token, Claude Sonnet 4.6 input costs 150% of Claude Sonnet 5.5's, and output costs 150%. Sources: Anthropic pricing, models overview.
What typical calls cost
Token counts per call are assumptions (shown in each row); adjust them in the Claude pricing calculator. These two models use different tokenizers: the newer one (Claude Opus 4.7 and later) fits about 555k words in 1M tokens versus about 750k before (Anthropic models overview), so the same text is about 1.35x the tokens on the newer model. The last column applies that to Claude Sonnet 5.5's input and output.
| Workload (input / output tokens) | Claude Sonnet 4.6 | Claude Sonnet 5.5 | Claude Sonnet 5.5, same text (tokenizer-adjusted) |
|---|---|---|---|
| Chat assistant turn (2,000 / 500) | $0.01350 | $0.00900 | $0.01217 |
| RAG / document Q&A call (8,000 / 400) | $0.03000 | $0.02000 | $0.02703 |
| Extraction or classification (1,500 / 150) | $0.00675 | $0.00450 | $0.00608 |
| Agent / coding loop step (30,000 / 2,000) | $0.12000 | $0.08000 | $0.10811 |
| Long-document summarization (150,000 / 1,500) | $0.47250 | $0.31500 | $0.42569 |
| Translation (1,000 / 1,200) | $0.02100 | $0.01400 | $0.01892 |
Before you switch
- Run your own evaluation: price differences are certain, quality differences depend on your task.
- Prompt caching and the Batch API change the totals a lot; model them in the prompt caching calculator and the batch API savings calculator.
- Check the API differences for the newer model in Anthropic's migration guides before changing a production setting.
More: all comparisons · LLM API pricing · cost by workload