Pricing — Tier & Model Explorer

Theoretical spend across pricing tiers, by model, in USD or EUR

Controls

Token volumes are derived from your real ccusage data (CLI + cowork). Tier rates are applied as a flat % discount on the per-model standard cost. EUR rate is editable — defaults to 0.92 (typical mid-market 2025–2026). Verify against your bank's rate for actual conversion.

What If You'd Been On A Different Plan?

Compares your actual API spend against subscription plans, projected to a full month and full year. Uses your usage rate of . The cheapest viable option is highlighted; amber rows likely exceed plan limits for typical Claude Code workloads.

Plan / Billing Mode Cost / month Cost / year vs API Standard (yr) Fit for your workload

Token volume: ~25M cache reads + ~3M cache writes + ~250K output per month at current rate. Subscription plans meter by message count and time-windowed limits (e.g. "messages per 5h"), not by tokens — so high-volume Claude Code workloads typically exceed Pro and even Max 5× before reaching the equivalent dollar value. API billing has no such limits.

Pricing Tier Definitions

Verified against Anthropic API docs and claudelog.com/claude-code-pricing. Anthropic does not publish % volume discounts for the API — Pro/Max are fixed-fee subscriptions, not API tiers.

Tier Discount vs Standard Source / Eligibility
No Cache+ ~265%Hypothetical: workload re-billed without prompt caching (cache reads at full input price). Shows what caching saves you.
Standard0%Default Anthropic API pricing — what Claude Code is billed at. Includes automatic prompt caching (cache reads at 10% of input).
Priority Tier0% (same price)Anthropic's committed-throughput tier — same per-token price as Standard; pays for guaranteed capacity, not a discount.
Batch API−50%Anthropic Batch API — 50% off input + output (and cache types). Async, 24h SLA. Not used by Claude Code (interactive).

Subscription Plans (Fixed-fee alternatives)

Pro and Max are fixed monthly fees with bundled usage limits — not per-token API tiers. Compare against your scenario's API spend to see if a subscription would have been cheaper. Claude Code on a subscription uses bundled allowance instead of API billing.

Plan Price / month Models Usage

Per-Million-Token Rates (Standard, USD)

Model Input Output Cache Write (5m, 1.25×) Cache Write (1h, 2×) Cache Read (10%)
Opus 4.5 / 4.6 / 4.7$5.00$25.00$6.25$10.00$0.50
Opus 4.6 — >200K context$10.00$37.50$12.50$20.00$1.00
Sonnet 4.6$3.00$15.00$3.75$6.00$0.30
Haiku 4.5$1.00$5.00$1.25$2.00$0.10

Per-Model Cost by Pricing Tier

Total Cost by Tier

Cost by Model & Tier (Stacked)