Claude Sonnet 4 vs Sonnet 5 price: cheaper per token, closer per task
Short answer: Claude Sonnet 4, 4.5 and 4.6 cost $3 per million input tokens and $15 per million output tokens. Claude Sonnet 5 costs $2 and $10, a third less per token. Sonnet 5 uses a newer tokenizer that produces about 30% more tokens for the same text, so for the same work the list-price difference is about 13%, not 33%.
| Model | Input | Output | Cache read | Cache write 5m / 1h | Tokenizer |
|---|---|---|---|---|---|
| Claude Sonnet 4 / 4.5 / 4.6 | $3 | $15 | $0.30 | $3.75 / $6 | previous |
| Claude Sonnet 5 | $2 | $10 | $0.20 | $2.50 / $4 | newer |
USD per million tokens, standard tier, global routing.
Price per token is not price per task. Anthropic notes that Claude 4.7 and later models use a newer tokenizer that produces about 30% more tokens for the same text, depending on the content (pricing page). Output length and quality on your own tasks also change the result, so treat the numbers below as a scenario and test on your own prompts.
A worked scenario
At $6,000 a month on Sonnet 4.x, the same work at Sonnet 5 list prices with 30% more tokens would be about $6,000 × 2/3 × 1.3 ≈ $5,200. The difference, around $800 a month, depends on your content and output length, so measure it on your own prompts.
What to do before switching
- Run a sample of real prompts on both models and compare quality and token counts.
- Check whether your prompts are cached; cache reads are also cheaper on Sonnet 5 ($0.20 versus $0.30).
- Switch one service first and compare the next week's bill.
Check your own bill
Open your usage or cost CSV in the free checkup. It shows spend per model and, for Sonnet 4.x, a price scenario at Sonnet 5 prices with the tokenizer assumption spelled out.
Questions
How much cheaper is Claude Sonnet 5 than Sonnet 4?
Per token, a third cheaper: $2 versus $3 per million input tokens and $10 versus $15 per million output tokens. Sonnet 5 uses a newer tokenizer that produces about 30% more tokens for the same text, so for the same work the list-price difference is about 13%.
Should I switch from Sonnet 4 to Sonnet 5?
Decide on your own evaluation: compare quality and token counts on a sample of real prompts. The price difference per task is smaller than the per-token prices suggest.
List prices checked 27 September 2026 on the official Anthropic and OpenAI price pages. Prices change; check the source before making decisions. Not affiliated with Anthropic or OpenAI. © 2026 Valeriy Danilov. AI Spend Doctor by AgentBridge Labs.