AI cost · Anthropic
Anthropic API cost: what you actually pay
Usage-based tokens by model. Watch the Opus premium — and that Opus 4.7’s tokenizer emits ~35% more tokens for the same text, quietly inflating bills at an unchanged per-token rate.
Public pricing
Token rates (USD per 1M tokens)
| Model | Input | Output | Cache read |
|---|---|---|---|
| Claude Opus 4.7New tokenizer emits ~35% more tokens for the same text. | $5 | $25 | $0.5 |
| Claude Sonnet 4.6 | $3 | $15 | $0.3 |
| Claude Haiku 4.5 | $1 | $5 | $0.1 |
Prompt caching cuts repeated-context reads to roughly a tenth of input price (see cache-read rates). Model choice dominates cost.
Last updated: 2026-07-22. Sourced from Anthropic pricing (official) ↗, Claude API rates (2026) ↗.
Where the real Anthropic API cost hides
The sticker is the smallest part of the story. For Anthropic API, these are the line items that quietly inflate the bill:
- Opus tokenizer inflation raising real cost at a flat per-token rate
- Prompt caching left off (cache reads are ~10× cheaper than fresh input)
- Opus used on tasks Sonnet or Haiku would handle
What an audit finds on Anthropic API
Opus on workloads Sonnet handles, caching disabled, and silent token growth from the new tokenizer inflating month-over-month spend.
Anthropic API pricing FAQ
How much does the Claude API cost?
Per million tokens: Opus 4.7 $5 in / $25 out, Sonnet 4.6 $3 / $15, Haiku 4.5 $1 / $5, with cache reads at roughly a tenth of input.
How do I cut Claude API cost?
Turn on prompt caching, drop from Opus to Sonnet/Haiku where quality allows, and watch tokenizer-driven token growth on Opus 4.7.
What is Anthropic API really costing you?
Spendassay connects your Anthropic API spend to actual usage and engineering output — a proof-level AI cost report with the wasted dollars named. Neutral across every AI vendor.
Free · read-only · no card
Other api platforms pricing
More: the complete AI cost management guide · every tool, with prices · cost per outcome · compare platforms · how we measure