Verified 2026-02-11 · sourced from Anthropic

100K Claude Opus 4.6 Tokens — Cost Breakdown

Use this guide to benchmark budgets for 100,000 tokens. Standard pricing is $5.00 per million input tokens and $25.00 per million output tokens. Cached input, when available, reduces costs significantly for repeated contexts.

Scenario breakdown

Cost estimates for different input/output distributions using 100K total tokens.

ScenarioTokens inTokens outStandard costCached cost
Balanced conversation
50% input · 50% output
50,00050,000$1.50$1.27
Input-heavy workflow
80% input · 20% output
80,00020,000$0.900$0.540
Generation heavy
30% input · 70% output
30,00070,000$1.90$1.77
Cached system prompt
90% cached input · 10% fresh output
90,00010,000$0.700$0.295

Workload multipliers

Convert 100K tokens into daily and monthly run-rate budgets.

ProfileRuns/dayTokens/dayDaily costMonthly costCached dailyCached monthly
Single workload1100,000$1.50$45.00$0.295$8.85
Daily batch (10 runs)101,000,000$15.00$450.00$2.95$88.50
Team workload (100 runs)10010,000,000$150.00$4500.00$29.50$885.00

Frequently asked questions

What is the standard cost of 100K Claude Opus 4.6 tokens?

100K tokens in a 50/50 conversation mix cost roughly $1.50 at the published Anthropic rates.

What happens if cached input is enabled?

With cached contexts, the same 100K tokens drop to $1.27 because input costs fall to $0.500 per million tokens.

How many requests does 100K tokens cover?

If your prompts average around 3,000 tokens per call, 100K total tokens cover about 33 requests.

How fresh is the pricing information?

Prices are taken from https://docs.claude.com/en/docs/about-claude/pricing and were last verified on 2026-02-11. models.json keeps this guide in sync with upstream changes.

Related resources