Claude Pricing Explained (2026)

Claude pricing is split across consumer plans (Pro, Max), API tiers (Opus, Sonnet, Haiku), and team / enterprise plans. Here’s every relevant 2026 number, plus when each tier is actually worth it.

Consumer plans

• Free — Claude Sonnet 4.6 with daily message cap (~10-30 depending on model load). Adequate for casual use.
• Pro ($20/mo) — 5x usage vs free, Opus 4.7 access, Claude Code, Projects, prompt caching in conversations. Sweet spot for most paying users.
• Max 5x ($100/mo) — 5x usage above Pro. Right when Pro caps you out daily.
• Max 20x ($200/mo) — 20x Pro usage. Right for heavy Claude Code users running long agentic tasks.

Team / Enterprise

• Team ($25-30/seat/mo) — min 5 seats, admin controls, shared Projects, central billing.
• Enterprise — SSO, SCIM, audit logs, expanded usage, custom contracts. Pricing is “contact sales.”

API pricing (per 1M tokens)

• Claude Opus 4.7: $15 input / $75 output. Cache write $18.75; cache read $1.50.
• Claude Sonnet 4.6: $3 / $15. Cache write $3.75; cache read $0.30.
• Claude Haiku 4.5: $0.80 / $4. Cache read $0.08.
• Batch API (50% off): async, 24h SLA. Same models.

The savings levers

• Prompt caching — 90% off cached input. Use it always.
• Batch API — 50% off when latency doesn’t matter.
• Sonnet over Opus — 5x cheaper at 95% the quality.
• Haiku for routine — 4x cheaper than Sonnet for autocomplete / classification.

When each tier is worth it

• Free: casual users testing what Claude is.
• Pro $20: default for any serious daily user.
• Max $100: if you hit Pro caps weekly. Common for Claude Code power-users.
• API: when building any product or running automated workflows. Start Tier 1 ($5 funded), grow with usage.

Compare to other plans: ChatGPT Plus vs Claude Pro. Run the API math: cost calculator, cache savings.