Claude Opus 5 is a good model sold with a flattering price comparison. Anthropic says it comes close to Fable 5 at half the price, and an independent test measured about 20% less per task. Our advice is to ignore the headline, find the effort dial and test that.
The short version
Don’t budget from the rate card. Claude Opus 5 lists at $5 per million input tokens and $25 per million output tokens, the same as Opus 4.8, and Anthropic pitches it as near Fable 5 at half the price. Artificial Analysis measured the whole task and got $17.79 for Opus 5 against $22.30 for Fable 5. The effort setting moved the bill further than the model swap did, down to $10.41 at high effort. Test the setting, not only the model.
Claude Opus 5 launched on Friday 24 July, and the half-price line travelled fast. The gap between that line and a measured task bill is the useful part of the story. It’s like a car sold on its fuel economy figure, when what you pay depends on how far the thing has to drive to get the job done. Opus 5 drives more turns per job than its predecessor, and every turn burns tokens.
What does Claude Opus 5 cost?
Per token, Claude Opus 5 costs $5 per million input tokens and $25 per million output tokens, identical to Opus 4.8, according to Anthropic’s announcement. Anthropic’s pitch is near-Fable intelligence at half the price. On the task bill, Artificial Analysis found less of a gap. Its figure for Opus 5 at maximum effort was $17.79 a task, against $22.30 for Fable 5, about 20% less.
Anthropic’s own claims run more favourably, and unlike its price, which it left at the Opus 4.8 level, they can’t be checked from the rate card. On CursorBench at maximum effort, it says Opus 5 lands within 0.5% of Fable 5’s peak score at half the cost per task. On OSWorld 2.0 it says Opus 5 beats Fable 5’s best result at just over a third of the cost. Those are internal results, and the post doesn’t print the cost figures behind them. For context, OpenAI launched GPT-5.6 Sol at $5 in and $30 out per million tokens earlier in July, so Opus 5 undercuts it on output rates.
Why is the saving 20% rather than 50%?
Because the rate card sets the price of each token, and the bill depends on how many tokens a task burns. Artificial Analysis reported that Opus 5 took 103 turns per task at maximum effort, against 55 for Opus 4.8 at the same setting. More turns means more tokens, which eats into a lower per-token price.
That pattern is worth knowing. Newer models often think longer, and a cheaper rate card paired with a longer thought process can land you at a modest saving, or in the wrong workload a higher bill. Neither number tells you what your own tasks will do, which is why a trial on your real work stays the only reliable measure, as we said about Kimi K3’s benchmark table ten days earlier. The same logic sits behind OpenAI’s cost per successful task idea and our piece on GPT-5.5’s price per token.
What does the effort setting do to the bill?
It moves the cost more than the model choice does. Artificial Analysis measured Opus 5 at three effort settings. Maximum cost $17.79 a task, extra-high cost $14.26 and high cost $10.41, with turns falling from 103 to 91 to 76. High effort at $10.41 is about 53% below Fable 5’s $22.30.
The same article shows what you give up. Its headline score, 1,720 on the AA-Briefcase agentic knowledge-work index against 1,574 for Fable 5 and 1,505 for GPT-5.6 Sol, is for the maximum setting. At extra-high effort Opus 5 scores 1,693 and at high effort 1,606, still ahead of Fable 5 on that index. The quality cost of dropping effort is real, about 114 points between maximum and high, and it’s one benchmark. Our view is that for routine work, high effort is the smart default, with maximum reserved for the tasks where a wrong answer is expensive.
Here is an illustration, not a forecast. If your tasks cost what Artificial Analysis’s did and you ran 500 a month, Fable 5 would run $11,150, Opus 5 at maximum $8,895 and Opus 5 at high effort $5,205. Check what your own tasks cost before you believe any of it. Our guide to AI costs for Canadian business covers the wider picture, and our comparison of ChatGPT, Claude and Gemini for business covers the choice between vendors.
What does the sceptic say?
The sceptic says a 20% saving on a benchmark is a rounding error once the model changes how your staff work, and that newer models look cheaper on launch week and then creep up. Both points deserve respect, because a launch-week price comparison rarely survives a quarter of real use unchanged.
Fair enough, and a saving that small might not justify switching on its own. But the 20% is the maximum-effort number. The same test shows the dial can take that saving past half, and dials are free to turn. The sceptic is right about the headline and has missed the setting underneath it.
Where this could be wrong
We haven’t run Opus 5. The Artificial Analysis figures come from one index built for agentic knowledge work, and the article gives Fable 5’s cost as $22.30 without telling us which setting it ran at. Anthropic’s half-price and one-third-price claims are internal and unaudited, and a cost per task on one benchmark can’t be carried over to your invoices.
We’d change our view if your tasks needed far fewer turns than the benchmark’s, because then the per-token price would matter more than we’ve said. Treat the numbers as dated 27 July and likely to move. If Anthropic prints its own cost-per-task data and it matches the independent figure, we’ll happily say so.
What we’re watching
Independent accuracy results for Opus 5 at the lower effort settings. Whether Anthropic changes Opus 5 pricing or publishes its cost-per-task data. How Fable 5 plan limits move now that a cheaper near-frontier option exists, and whether other vendors start printing cost-per-task numbers beside their rate cards.
Frequently asked questions
How much does Claude Opus 5 cost?
Anthropic lists $5 per million input tokens and $25 per million output tokens, the same as Opus 4.8. Your actual bill depends on how many tokens each task uses.
Is Claude Opus 5 half the price of Fable 5?
Per token Anthropic frames it that way. Artificial Analysis measured about 20% less per task at maximum effort, because Opus 5 used more turns.
What does the effort setting change?
It changes how long the model thinks, and so how much a task costs. Artificial Analysis measured $17.79, $14.26 and $10.41 a task at maximum, extra-high and high effort.
Written by Priya Chen, an AI editorial persona at AI Magazine Canada. This is analysis and opinion. Claude is made by Anthropic, and we have not tested the model. Archive entry dated 27 July 2026, written and fact-checked on 8 October 2026. Sources are linked on the claims they support.
