AI API Cost Calculator

Calculate and Compare
AI API Costs

Estimate your AI spending, compare model pricing, and understand what your real-world workloads actually cost.

Calculate Your Costs Compare Models
11Pricing guides
8Models in calculator
Oct 2026Prices verified

Claude Opus 5.5 API Pricing: Per-Token Costs and Workload Math

How much does the Claude Opus 5.5 API cost per 1M tokens?

Claude Opus 5.5 costs $4.00 per 1M input tokens, $0.20 per 1M cached input tokens, and $20.00 per 1M output tokens. Cache reads run at 5% of the input price — half the ratio of most Claude models, and the deepest cut in this release. Last verified: October 2026, against Anthropic's official pricing documentation.

The Full Price Table

Straight from Anthropic's published rate card (standard tier):

Token typePrice per 1M tokens
Input$4.00
Cached input (reads)$0.20
Cache writes (5-minute)$5.00
Cache writes (1-hour)$8.00
Output$20.00

The cache-read line is the story of this release. On most Claude models a cache hit costs 10% of the input price; on Opus 5.5 it's 5%. Anthropic kept the standard write multipliers — 1.25x input for the default 5-minute cache, 2x for the 1-hour option — so the first call that establishes a new cache prefix costs slightly more than an uncached call, and every reuse after that is nearly free. The 1-hour cache costs double the 5-minute one; buy it only when your prefix genuinely sits idle between half-hour gaps.

What Changed vs Opus 5

Opus 5.5 replaces Opus 5 at the top of Anthropic's lineup (launched September 22, 2026, API ID claude-opus-5-5). Every line of the rate card moved:

Rate (per 1M tokens)Opus 5Opus 5.5Change
Input$5.00$4.00-20%
Output$25.00$20.00-20%
Cache read$0.50$0.20-60%
Cache write (5 min)$6.25$5.00-20%
Cache write (1 hr)$10.00$8.00-20%

Every line moved 20% except one: cache reads fell 60%. That's deliberate — in agentic coding work the model re-reads the same cached context on every turn, and Anthropic says cache reads dominate those bills. Anthropic claims the combination of lower rates and lower token use cuts typical workload cost by about 40% (their measurement — treat vendor arithmetic accordingly).

The Workload Math: $195.36/Month

Headline rates don't pay invoices; workloads do. Run Opus 5.5 through the reference workload — 1,100 tasks per month, each with 12K fresh input, 48K cached input, and 6K output tokens:

  • Fresh input: 12,000 x $4.00/1M = $0.0480 per task
  • Cached input: 48,000 x $0.20/1M = $0.0096 per task
  • Output: 6,000 x $20.00/1M = $0.1200 per task
  • Per task: $0.1776 → per month (x1,100): $195.36

Notice the shape: output is two-thirds of the per-task cost, and the 48K cached tokens — four times the fresh input volume — cost a fifth of it. That's the 5% cache-read rate earning its keep. Without caching, those 48K tokens at full price would cost $0.1920 per task instead of $0.0096 — the cache line is doing the heaviest lifting on this bill.

How It Compares on the Same Workload

Same 1,100-task workload, verified current prices:

ModelMonthly costvs GPT-6.1 Sol
GPT-6.1 Sol$97.68—
Claude Sonnet 5$102.96+5.4%
Claude Opus 5.5$195.36+100%
Gemini 4 Argon (intro, thru Nov 30)$198.00+103%
GPT-5.6 Sol (promo, thru Nov 21)$205.92+111%
Gemini 4 Argon (from Dec 1)$396.00+305%

Two honest caveats. First, these are standard-tier rates — the fine print moves them. Second, GPT-5.6 Sol's promo expires November 21 with no published successor price (pre-promo list was $5/$30, which would put it at $290.40/month). Argon publishes no cache rate, so its cached tokens are modeled at the full input rate — on cache-heavy workloads the gap is wider than shown.

The uncomfortable truth in the table: Opus 5.5 costs exactly twice what GPT-6.1 Sol costs on the same workload. You're not buying Opus 5.5 to save money — you're buying Anthropic's flagship for the work where its capability earns the premium. If the workload doesn't need the top tier, Sonnet 5 at $102.96 does the same shape of math for nearly half.

The Fine Print That Moves Your Bill

  • Batch API: 50% off both directions. Asynchronous jobs completed within 24 hours run at $2/$10. The discount stacks with caching — a batched cache hit costs $0.10/1M. On the reference workload, batch mode drops the bill to roughly $97.68/month, exactly matching GPT-6.1 Sol's standard price.
  • No long-context surcharge. The full 1M-token context window bills at the standard per-token rate — no repricing tier like OpenAI's 272K threshold. A 900K-token request costs the same per token as a short one. This is a structural difference from OpenAI and Google.
  • Fast mode: $8/$40 (2x), research preview. Up to 2.5x faster output, available only on Anthropic's first-party API — not AWS, Google Cloud, or Microsoft Foundry — and it can't be combined with batch.
  • Thinking tokens bill as output. Opus 5.5's adaptive thinking is always on and can't be disabled; reasoning tokens count toward output at $20/1M whether or not you read them. Budget for the thinking, not just the answer.
  • Context window: 1M tokens in, 128K max output per synchronous reply (300K via the Batch API behind a beta header). Knowledge cutoff June 2026.
  • Retirement floor: Anthropic says Opus 5.5 won't retire sooner than September 22, 2027 — safe to pin claude-opus-5-5 in production code.

FAQ

How much does Opus 5.5 cost per request?

$0.1776 on the reference workload (12K fresh input + 48K cached input + 6K output). A simple 1K-input / 1K-output request costs about $0.024.

Is Opus 5.5 cheaper than GPT-6.1 Sol?

No — it's exactly double: $195.36/month vs $97.68/month on the same workload. Opus 5.5 is Anthropic's flagship; GPT-6.1 Sol is the value pick. Buy Opus 5.5 for capability, not for the invoice.

Does Opus 5.5 have a promo price that expires?

No. The $4/$20 rate is the standard list price with no published expiry — unlike GPT-5.6 Sol's promo (ends November 21, 2026) and Gemini 4 Argon's intro rate (doubles December 1, 2026). The price is stable; whether it's worth it is the actual question.

What's the cheapest way to run Opus 5.5?

Combine Batch API (50% off) with prompt caching: fresh input at $2.00, cache hits at $0.10, output at $10.00 per 1M. On the reference workload that's roughly $97.68/month — half the standard bill. The catch is latency: batch jobs are asynchronous, so this works for pipelines and evals, not interactive products.

The Bottom Line

Opus 5.5 is the most interesting price cut Anthropic has shipped: not the 20% off the headline, but the 60% off cache reads, aimed at the agentic coding loops that re-read the same context thousands of times. At $195.36/month on the reference workload it's still twice GPT-6.1 Sol — the flagship premium is real. But it's a stable, published $4/$20 with no promo cliff and no long-context tax, and batch mode halves everything. If your workload needs the top tier, this is the cheapest it's ever been to run it.


 

Popular posts from this blog

Argon vs Wonder: Which AI Should Developers Build On?

Image
Last updated: October 2026 Two flagship models, two very different price tags — and one of the prices is about to double. If you're choosing a foundation model to build on right now, the decision isn't just about capability. It's about timing. Here's the honest comparison: what each model costs, what the benchmarks actually say, and which one to pick depending on what you're optimizing for. The Pricing: Read the Dates Gemini 4 Argon is on introductory pricing: $2.00 per 1M input tokens / $10.00 per 1M output tokens . That intro rate runs through November 30, 2026 . From December 1 , the price doubles to $4.00 / $20.00 . Muse Wonder is priced at $4.00 / $20.00 per 1M tokens — no intro period, no scheduled change. So today, Argon costs half what Wonder does. On December 1, they cost exactly the same. That convergence matters more than it looks. Every cost comparison you run today has a 6-week expiry date. If you're doing the math — and you should be — do...
Read more

GPT-5.6 API Pricing: The November 21 Deadline

Image
Last updated: October 2026 If you're running GPT-5.6 through the API right now, you're on a promotional price. And that promotion has an expiry date: November 21, 2026 . OpenAI's pricing page states the current rate is "available at least through November 21, 2026." Read that carefully. "At least through" — not "until." The promo could extend. It could also end exactly on that date. OpenAI has published no pricing for what comes after. This post lays out exactly what the pricing is today, what it costs on a realistic workload, and how to budget for the unknown that follows November 21. The Current Pricing, Exactly GPT-5.6 is available in several variants. The one most developers are using — the flagship reasoning model, referred to here as GPT-5.6 Sol — is currently priced at: Input: $4.00 per 1M tokens Cached input: $0.40 per 1M tokens Output: $20.00 per 1M tokens That is promotional pricing. Before the promo, announced August 21, ...
Read more

Prompt Caching: The 90% Discount You're Missing

Last updated: October 2026 There is a 90% discount sitting in your API bill, and most developers are either not using it or using it in a way that doesn't work. Prompt caching — paying a tenth of the input price for tokens the provider has already seen — is the highest-leverage cost optimization available on modern LLM APIs. Here's how it actually works and how to get the savings. How Prompt Caching Actually Works The idea is simple. When you send a request, the provider processes your input tokens through the model — that's the expensive computation you're paying for. But if your next request starts with the exact same tokens — the same system prompt, the same repo context, the same few-shot examples — the provider doesn't need to recompute all of that. It kept the intermediate results cached. So it charges you a fraction of the price for those repeated tokens. The key phrase is stable prefix . Caching works from the start of your prompt forward: the provider ...
Read more