Claude Opus 5.5 pricing fell 20% on the list, and the headline will tempt you to bank the saving. Our view is that the effort dial moves your bill by far more than the cut does, so set the default to medium and make anyone who wants max say why.
The short version
- Anthropic’s list rates fell to $4 input and $20 output per million tokens, from $5 and $25 on Opus 5. That’s a 20% cut.
- Anthropic says typical workloads will cost 40% less. That’s its claim, it names no test, and we’d believe it when your own invoice agrees.
- Third-party data prices a single task anywhere from $0.55 to $5.98 depending on effort. Pick the dial deliberately, because it matters more than the price list.
Follow the money from the top. Anthropic’s launch page lists Opus 5.5 at $4 and $20, against $5 and $25 for Opus 5, with cache reads at $0.20 versus $0.50. It says the model uses fewer tokens per task, which is where the 40% would come from. Anthropic’s pricing docs show the same rates.
What changed in Claude Opus 5.5 pricing?
The price per token fell 20%, and Anthropic says the cost of a typical job falls further because the model needs fewer tokens to finish it. The first number sits on a price list. The second is an estimate with no workload attached, so treat it as a hypothesis until your invoice says otherwise.
The fine detail is worth a look. Fast mode, up to 2.5x speed, costs $8 and $40, double the standard rates. Batch work is half price, at $2 and $10 per Anthropic’s docs. Subscribers on paid Claude plans get higher five-hour limits and a saveable rate-limit reset, according to the launch page, though it gives no figures. On an API plan, the list price is your number. On a seat plan, the limits are.
Why does the effort setting matter more than the price cut?
Because the amount of thinking a model does scales your token count, and tokens are what you pay for. On Artificial Analysis data reported by Digital Applied, Opus 5.5 cost $0.55 per task on low effort, $1.34 on medium, $3.46 on xhigh and $5.98 on max.
Max effort used 119,200 output tokens per task against 25,700 at medium. That’s about 4.6 times the tokens and 4.5 times the cost, while the intelligence index score moved from 51.2 to 57.6. Paying 4.5x for roughly 12% more score isn’t automatically wrong, and some tasks justify it. Most email drafts and spreadsheet clean-ups don’t. A 20% list cut is worth having, but the dial is worth more than four times as much.
Does GPT-6 Sol change the maths?
It changes the comparison, though less than OpenAI’s page suggests. OpenAI’s announcement puts GPT-6 Sol at $2 and $10 and Luna at $0.10 and $0.50, each 50% below earlier promotional prices. The page carries no date, and Sol and Luna were available in ChatGPT Work and Codex, not regular Chat.
OpenAI’s own cost-per-task claims compare Sol with Opus 5, not Opus 5.5. The company says Sol scores 33.2% on AutomationBench at $0.27 per task, against 26.9% for Opus 5 at about 11.1 times the cost. That’s a vendor claim on one benchmark, and cheap per token can still be expensive per job. We made the same point about Claude Opus 5’s cost per task, and about the previous price cut in our Opus 4.8 pricing piece.
Show me the invoice
Take a team running 2,000 AI tasks a month, a mix of document drafts, data clean-up and research summaries. Using Artificial Analysis’s per-task costs as a stand-in, medium effort costs about $2,680 a month ($1.34 times 2,000). Max effort on everything costs about $11,960. That gap of roughly $9,300 a month comes from a drop-down menu, not a contract, and no finance team would sign off on it if it arrived as a line item.
Those per-task figures come from Artificial Analysis’s test set, not your work, so the totals are an illustration. The rule that survives is simple. Set medium as the team default. Allow high where a wrong answer costs real money. Allow max only with a named reason. Then track cost per accepted output, not cost per request, because a rejected draft still costs you the tokens. We cover that habit in our piece on cost per successful task and in our guide to AI costs for Canadian businesses. The same logic applied when token prices fell and a budget still blew out.
Where this could be wrong
Our case rests on the cost-per-task figures in Digital Applied’s report of Artificial Analysis data published on 22 September, and Artificial Analysis may revise them. They also show Opus 5.5 running with a fallback enabled, which means some requests that tripped a safeguard were answered by another Claude model. Artificial Analysis’s own test setup also scored Opus 5.5 below Anthropic’s figure on Terminal-Bench 4.0, so vendor benchmarks and independent ones differ.
One more caution cuts against cheap defaults. The same report says Opus 5.5 gave a wrong answer 68.4% of the time when it didn’t know, where two other models more often declined. Cost per task isn’t cost per correct task. If your documents are the kind where max effort catches errors medium misses, our advice flips for that work. Only your own sample will tell you.
The sceptic’s best case
The sceptic says nobody in a 40-person firm will tune an effort dial, so default behaviour wins, and the vendor sets the default. That’s right, and it’s the reason to look. If your tooling defaults to medium, the dial matters only when someone pushes it up, and you should ask who is doing that. If it defaults to high or max, you’re paying the premium without noticing.
We read the sceptic’s point as a reason to check the setting once, then leave it alone. One check beats a quarterly argument about the bill.
What to watch
- Whether Anthropic publishes the workload behind its 40% claim.
- Whether Artificial Analysis revises its Opus 5.5 numbers.
- How GPT-6 Sol and Luna prices hold once they reach regular Chat.
- Whether seat-plan limits, rather than API prices, become the real constraint.
Frequently asked questions
How much does Claude Opus 5.5 cost?
Anthropic lists it at $4 per million input tokens and $20 per million output tokens, down from $5 and $25 for Opus 5. Fast mode costs $8 and $40, and batch jobs cost $2 and $10.
Is Opus 5.5 really 40% cheaper than Opus 5?
That is Anthropic’s claim for typical workloads, based on the model using fewer tokens per task. The list price cut is 20%. Check the 40% against your own invoice.
What effort setting should a small business use?
Medium is a sensible default, since Artificial Analysis data shows it at $1.34 per task against $5.98 at max. Use higher settings only where a wrong answer is costly.
Written by David Okafor, an AI editorial persona at AI Magazine Canada. This is analysis and opinion. Archive entry dated 23 September 2026, written and fact-checked on 8 October 2026. Sources are linked on the claims they support.
