Cafiyn Pulse
Anthropic · Verified 2026-09-05

Claude Sonnet 5 pricing, broken down.

Anthropic's mid-tier flagship. Here is what it actually costs per token, and what that means for a real monthly bill.

Claude Sonnet 5
Anthropic · premium tier
Verify against Anthropic's pricing page ↗
$2.00
per 1M input tokens
$10.00
per 1M output tokens
A typical 1,200-in / 400-out token request costs about $0.0064, or roughly $6.40 per 1,000 requests.

Claude Sonnet 5 sits in Anthropic's lineup as the model most production apps default to: strong reasoning, fast enough for interactive use, and priced well below Opus while comfortably ahead of Haiku in capability. It is the model most teams should reach for first, then downgrade specific call sites to Haiku once they know which ones do not need the extra reasoning.

Pricing is asymmetric between input and output, output tokens cost five times more than input tokens, which matters more than it first appears. A summarization task (long input, short output) is cheap. A generation task (short input, long output) is not. Budget from your actual input:output ratio, not just total token count.

Best for
  • Production chat and support assistants that need reliable reasoning at interactive speed
  • Code generation and review where correctness matters more than raw speed
  • RAG and agentic workflows with multi-step tool use
Head to head

Claude Sonnet 5 vs the alternatives.

ModelInput / 1MOutput / 1MSample request
Claude Sonnet 5$2.00$10.00$0.0064
GPT-5$1.25$10.00$0.0055
Gemini 3.1 Pro$2.00$12.00$0.0072
Sample request assumes 1,200 input and 400 output tokens, a typical chat exchange. Your actual ratio will differ, run your own numbers in the calculator below.
FAQ

Common questions.

How much does Claude Sonnet 5 cost per 1,000 requests?

It depends entirely on tokens per request, there is no fixed per-request price. A typical chat exchange (1,200 input / 400 output tokens) costs roughly $0.0064 per request, so 1,000 requests would run about $6.40. Use the calculator above with your own token counts for an accurate number.

Is Claude Sonnet 5 cheaper than GPT-5?

They are close. Sonnet 5 runs $2/million input and $10/million output; GPT-5 runs $1.25/million input and $10/million output, so GPT-5 is cheaper on input and identical on output. The gap is usually smaller than the difference model choice makes on output quality and task success rate.

Does Claude Sonnet 5 have a free tier?

Anthropic offers limited free usage through the Claude chat apps, but the API used for building products is pay-as-you-go with no free monthly token allowance as of this writing. Check Anthropic's pricing page for current promotional credits for new accounts.

More models

Browse other pricing pages.

Open the tool.

Live math against Anthropic's current pricing plus every other provider in the roster.

Compare Claude Sonnet 5 on my usage