Provider

Anthropic API pricing

Maker of Claude. Strongest current lineup for long-horizon agentic work and coding, with a 1M-token context window across the current models.

Anthropic model prices per million tokens
Relative cost
Claude Opus 5Anthropic$5.00$25.001M
Claude Opus 4.8Anthropic$5.00$25.001M
Claude Fable 5Anthropic$10.00$50.001M
Claude Sonnet 5Anthropic · Introductory pricing to 2026-08-31$2.00$3.00$10.00$15.001M
Claude Sonnet 4.6Anthropic$3.00$15.001M
Claude Haiku 4.5Anthropic$1.00$5.00200K

6 models · verified 24 June 2026 · official pricing ↗ · docs ↗

Worth knowing

  • The cheapest entry point is Claude Haiku 4.5 at $1.00 per million input tokens, with a 200K context window.
  • Claude Fable 5 — Thinking is always on and cannot be disabled. Requires 30-day data retention, so it is unavailable to zero-data-retention organisations.
  • Claude Opus 5 — Half the price of Fable 5 with thinking on by default. Caches from 512 tokens — the lowest minimum of any current model, so short prompts cache that would not elsewhere.
  • Claude Opus 4.8 — Same price as Opus 5. Kept current as the recommended fallback target when Opus 5 declines a request.
  • Claude Sonnet 5 — Near-frontier quality on coding and agentic work at Sonnet cost. Uses a newer tokenizer that produces roughly 30% more tokens than Sonnet 4.6 for the same text — re-measure before comparing bills.
  • Claude Sonnet 4.6 — The previous Sonnet. Same sticker price as Sonnet 5 but an older tokenizer, so a given prompt costs less in practice.
  • Claude Haiku 4.5 — The cheapest and fastest current Claude. Note the smaller context window and output cap.

Price a workload on these modelsCompare against other providers