Claude Cost Calculator

Estimate Anthropic Claude API token spend using current model, cache-read, and Batch API rates.

Inputs

Model

Choose the model tier you plan to use in production.

Input tokens / request

Average prompt size per API call — system prompt + context + user message. Typical ranges: simple chat 500–2,000 · RAG chatbot 2,000–8,000 · document analysis 10,000–100,000. A page of text ≈ 750 tokens.

Output tokens / request

Average response length per API call. Typical ranges: short answer 200–500 · paragraph 500–1,000 · long-form 1,000–4,000 tokens.

Monthly requests

Total API calls per month across all users and jobs. 10 users × 3 sessions/day × 10 calls/session × 30 days = 9,000 requests/month.

Cache hit ratio: 0.00

Fraction of input tokens billed as cached reads. Cache writes and storage are not included. Use 0 if caching is disabled.

Apply published Batch API rate(Uses provider/model-specific published Batch token rates. If none are published, standard rates remain in effect.)

Estimated yearly cost

$2,700.00/year

((2,000 fresh input × $5 + 0 cached-read input × $0.5 + 500 output × $25) / 1M × 10,000 req) × 12

Decision Summary

Best move

Evaluate DeepSeek DeepSeek V4 Flash — potential $2,649.60/yr saving

Expected savings

$2,649.60 /yr (98.1% reduction)

Watch out

Validate quality, latency, and reliability on your own workload before changing production traffic.

Next step

Optimize with AI Cost Optimizer

Evaluate DeepSeek DeepSeek V4 Flash — potential $2,649.60/yr saving

At published token rates, DeepSeek DeepSeek V4 Flash handles the same token volume at $50.40/yr — 98.1% less than Claude Claude Opus 5. This is a price-only comparison; validate quality, latency, tool support, and regional pricing before switching.

$2,649.60

saved / year (98.1%)

Prompt Cache not enabledEnable Prompt Cache

Enabling Prompt Cache with a 60% hit ratio could save $648.00/yr (24% reduction). Fix your system prompt at the start of every request and reuse it across calls.

$648.00

saved / year

Monthly cost

$225.00/month

Daily cost

$7.50/day

Cost per request

$0.022500

Model

Claude Opus 5

Input rate

$5/MTok

Output rate

$25/MTok

Cost breakdown

Item	Monthly	Yearly
Fresh input (2,000 tok/req × 10,000 req × $5/MTok)	$100.00	$1,200.00
Output spend (500 tok/req × 10,000 req × $25/MTok)	$125.00	$1,500.00
Total monthly spend	$225.00	$2,700.00

Comparison

Option	Monthly	Yearly
OpenAI GPT-5	$75.00	$900.00
Claude Claude Opus 5current	$225.00	$2,700.00
Gemini Gemini 2.5 Pro	$75.00	$900.00
DeepSeek DeepSeek V4 Flashcheapest	$4.20	$50.40

Pricing sources

Last verified 2026-08-03 · Claude official pricing platform.claude.com/docs/en/about-claude/pricing

Industry Benchmark

Output price vs. peer average ($/1M tokens)Industry avg: 11.32 $/1M

You are at the 110th percentile

Pricing sources

Last verified 2026-08-03 · developers.openai.com/api/docs/models/compare developers.openai.com/api/docs/models/compare · platform.claude.com/docs/en/about-claude/pricing platform.claude.com/docs/en/about-claude/pricing · ai.google.dev/gemini-api/docs/pricing ai.google.dev/gemini-api/docs/pricing · api-docs.deepseek.com/quick_start/pricing api-docs.deepseek.com/quick_start/pricing

Continue your analysis

Optimize with AI Cost Optimizer

See all savings opportunities ranked by impact.

Check SaaS pricing viability

Find the subscription price that covers your AI cost.

Calculate migration ROI

See payback period for switching to a cheaper model.

Trends & comparison

Trend

Comparison (monthly vs. yearly)

Official Anthropic rates used

Verified August 3, 2026: Claude Opus 5 and Opus 4.7 are $5/MTok input, $25 output, and $0.50 cached read. Sonnet 4.6 is $3/$15/$0.30; Haiku 4.5 is $1/$5/$0.10. Sonnet 5 is temporarily $2/$10/$0.20 through August 31, 2026.

A reproducible workload example

For 50,000 monthly calls with 800 input and 400 output tokens, token spend is $140 on Haiku 4.5, $420 on Sonnet 4.6, and $700 on Opus 5 before cache or Batch adjustments. These figures are direct token arithmetic, not performance claims.

Caching and Batch assumptions

Anthropic publishes separate cache-write and cache-read rates, and says Batch discounts can stack with prompt caching. This calculator models cached reads and the token-rate Batch multiplier, but excludes cache-write premiums, storage, tools, and other service charges.

Validate capability before routing

The calculator intentionally does not claim that a cheaper Claude tier has equivalent quality. Build an eval set from production tasks, establish pass thresholds, and route only the task classes that the lower-cost model handles reliably.

Frequently asked questions

What Claude prices does this calculator use?▾

Claude Opus 5 and Opus 4.7 use $5/MTok input and $25 output; Sonnet 4.6 uses $3/$15; Haiku 4.5 uses $1/$5. Sonnet 5 uses its published $2/$10 introductory rate through August 31, 2026.

Will Claude Sonnet 5 pricing change?▾

Anthropic says Sonnet 5 introductory pricing ends August 31, 2026 and becomes $3/MTok input and $15 output on September 1. The site must update the rate table on that date.

How is Claude prompt caching represented?▾

The cache slider represents cached reads at the published read rate. The estimate excludes the initial cache-write premium and any provider-specific charges, so it is a steady-state estimate rather than an invoice forecast.

Which Claude model should I choose?▾

Price alone cannot answer that. Run Opus, Sonnet, and Haiku against the same representative prompts and acceptance criteria, then compare total cost only among models that pass your quality and latency thresholds.

Related calculators

Same cluster

Compare verified first-party API prices on the same standard workload → AI Model Price Leaderboard