Anthropic

Claude Opus 4.6 pricing

Anthropic's flagship model. 1M context window and top-tier reasoning at a 5:1 output-to-input ratio.

Input

$5.00/ 1M tok

Output

$25.00/ 1M tok

Context window
1M
Max output
128K
Cached input
-
Verified
2026-04-06

Claude Opus 4.6 sits at the top of Anthropic's line-up and is the only model in the Claude family that ships with a 1M-token context window as standard. At $5 per 1M input and $25 per 1M output it's 3x cheaper than the legacy Opus 4.1 while matching it on output quality.

The 5:1 output-to-input ratio means the bulk of your bill lands on the response, not the prompt. A 10K-token prompt with a 2K-token reply runs about $0.10 - cheap in isolation, expensive at volume. For chat, summary, and routing, try Sonnet 4.6 first ($3 / $15) and reach for Opus only when there's a measurable capability gap on your task.

Calcis counts input tokens against Anthropic's own countTokens API - the same boundary Anthropic uses for billing - and predicts response length from the prompt before you send.

Estimate your cost on Claude Opus 4.6

Paste your prompt into the estimator, pick Claude Opus 4.6, and see the exact dollar cost - input tokens counted with the provider's own tokenizer, output tokens predicted by our regression model.

Open cost calculator →

Frequently asked

How much does Claude Opus 4.6 cost per request?
A 1,000-token prompt with a typical 500-token reply costs about $0.0175 ($0.005 input + $0.0125 output). Your exact cost depends on both sides - Calcis estimates both without guesswork.
Is Opus 4.6 more expensive than Sonnet 4.6?
Yes - about 1.7x on both input and output. Sonnet 4.6 runs $3/$15 per 1M tokens vs Opus's $5/$25. Both ship with a 1M context window, so pick on capability rather than context size.
Does Opus 4.6 have a long-context surcharge?
No. Unlike Gemini 2.5 Pro, Anthropic bills at flat per-token rates across the full 1M context window.
What tokenizer does Claude Opus 4.6 use?
Anthropic doesn't publish the tokenizer, so Calcis uses Anthropic's own countTokens API for the authoritative count. That's the same boundary Anthropic uses for billing.

Pricing verified 2026-04-06 from the provider's rate card.