Anthropic
Claude Opus 4.1 pricing
Legacy Opus. Still available at 2025 pricing - 3x more expensive than current Opus 4.6.
Input
$15.00/ 1M tok
Output
$75.00/ 1M tok
- Context window
- 200K
- Max output
- 32K
- Cached input
- -
- Verified
- 2026-04-06
Opus 4.1 was Anthropic's flagship through most of 2025 and is still available on the API at its original $15 / $75 per 1M tokens. That's 3x more expensive than the current Opus 4.6 ($5 / $25) for comparable reasoning quality on most tasks, so the only real reasons to stay on 4.1 are version pinning for audit reproducibility or a specific benchmark where 4.1 still outperforms the newer release.
A 10K-token prompt with a 2K-token reply runs about $0.30 on Opus 4.1 - versus $0.10 on Opus 4.6 for the same shape. At 10K requests a month that gap is $2,000, so audit your usage before leaving production code pinned here.
Calcis counts input tokens against Anthropic's countTokens API for the exact figure Anthropic uses to bill, and predicts response length from the prompt so you can see the full dollar number before you send.
Estimate your cost on Claude Opus 4.1
Paste your prompt into the estimator, pick Claude Opus 4.1, and see the exact dollar cost - input tokens counted with the provider's own tokenizer, output tokens predicted by our regression model.
Frequently asked
- Why is Opus 4.1 so much more expensive than 4.6?
- 4.1 was priced before Anthropic cut Opus rates for 4.5/4.6. They kept 4.1 on its original rate card rather than retroactively repricing. New deployments should default to 4.6 unless you need 4.1 specifically.
- How much does Opus 4.1 cost per request?
- A 1,000-token prompt with a 500-token reply costs about $0.0525 ($0.015 input + $0.0375 output). The same request on Opus 4.6 costs $0.0175 - 3x less.
- Should I migrate from Opus 4.1 to 4.6?
- Probably yes, unless you have reproducibility requirements. 4.6 is cheaper, has a larger context window (1M vs 200K), and performs at least as well on most tasks. Re-run your benchmarks before migrating production workflows.
- Does Opus 4.1 support prompt caching?
- Yes - Anthropic's prompt caching works on Opus 4.1 the same way as other Claude models. Cache reads are 0.1x the input rate (a 90% discount on repeated prefixes).
Pricing verified 2026-04-06 from the provider's rate card.