Anthropic › Claude Sonnet 5 › Pricing

Claude Sonnet 5 pricing

API list price by Anthropic · checked Oct 23, 2026
Benchmarks and alternatives →
$2 / $10per 1M tokens in / out
$0.20cache read · −90%
$1 / $5batch · −50%
$1.00per task at medium?
#39on OpenRouter by tokens · 578.2B / week

Price sheetper 1M tokens

Input$2list
Output$10
Cache read$0.20−90% vs input
Cache write · 5 min$2.50+25% vs input
Cache write · 1 h$4+100% vs input
Batch input$1−50%
Batch output$5−50%
Context · max output1M · 128Ktokens

Cost per task by effort?

low$0.5121.9
mediumAPI default$1.0030
high$1.7935.7
xhigh$2.8739
max$5.0941.3
per taskHumanity's Last Exam
Same price per token at every effort; the bill grows with reasoning tokens. max costs 5.1× medium for +11.3 points.

Where to buy6 providers · per 1M tokens

ProviderInputOutputContext
Amazon Bedrock$2$101M
Anthropiclab$2$101M
Claude Platform on AWS$2$101M
Google$2$101M
Azure$2.20$111M
DeepInfra$3$15—

In Anthropic's line-upper 1M in / out

Claude Haiku 5.5$0.10 / $0.50−95%
Claude Sonnet 5$2 / $10this model
Claude Sonnet 5.5$2 / $100%
Claude Opus 5.5$4 / $202×
Claude Fable 5.1$10 / $505×
Claude Mythos 5$10 / $505×
Claude Mythos 5.1$10 / $505×
<

Cheaper for the same workvs Sonnet 5 at medium

GPT-6.1 Sol max-28%score +85%$0.72
Claude Sonnet 5.5 high-11%score +67%$0.88
MiMo v2.6 Pro thinkingOpen-87%score +65%$0.13
GPT-6 Astra low-18%score +63%$0.82
DeepSeek V4.1 Flash 1017 maxOpen-74%score +61%$0.26

Pricing questions

How much does Claude Sonnet 5 cost?›

$2 per 1M input tokens and $10 per 1M output tokens. Cached input is $0.20. Batch requests cost $1 / $5. One Intelligence Index task costs about $1.00 at the default effort.

Is Claude Sonnet 5 free to use?›

There is no free API tier.

How much does prompt caching save?›

Cache reads cost $0.20 per 1M, 90% below the input price. Writing to the cache costs $2.50 per 1M (5m TTL).

Is Claude Sonnet 5 cheaper than Claude Sonnet 4.6?›

Yes: $2 / $10 against $3 / $15, 33% less per token.

How does effort change the price?›

The price per token is the same at every effort. Higher effort spends more reasoning tokens: a task costs $0.51 at low, $1.00 at medium, $1.79 at high, $2.87 at xhigh, $5.09 at max.

What is a cheaper alternative to Claude Sonnet 5?›

GPT-6.1 Sol at max effort costs -28% per task with a score +85% on Humanity's Last Exam.