1300 166 138 Tell us what you need
News / Claude

Claude Opus 5 lands as the new default for office AI

By Monika Tantau • Published 4 months ago
Claude Opus 5 lands as the new default for office AI

Key takeaways

• Anthropic released Claude Opus 5 on 24 July 2026, priced at $5 per million input tokens and $25 per million output tokens.
• It approaches the quality of the flagship Fable 5 in many tasks at half the price, and is now the default model on Claude Max.
• A new effort setting lets teams dial the model between low, medium and high to trade cost against capability.
• For Australian teams, the shift is about cheaper everyday automation rather than a jump in raw intelligence.

Most Read

Anthropic has released Claude Opus 5, a model it positions as its best value for everyday work, on 24 July. The company describes it as a thoughtful and proactive model that comes close to the frontier intelligence of its flagship Claude Fable 5 in many categories, but at half the price, according to Anthropic.

The headline for most businesses is not a new intelligence ceiling. It is cost. Opus 5 runs at $5 per million input tokens and $25 per million output tokens, the same rate as Opus 4.8, while doing noticeably more for that money. Anthropic expects it to become the default choice for a lot of day to day office work.

What Anthropic announced

Opus 5 is now the default model on Claude Max and the strongest option on Claude Pro, so most people on those plans will start using it without changing anything. It ships with an effort setting that lets you choose how hard the model works on a task, from low through medium to high, which is a direct lever on both quality and cost.

Two smaller updates arrived alongside it in beta. Developers can now change which tools the model has access to partway through a conversation without resetting its cached context, and requests that get flagged by safety checks can be routed automatically to another model rather than simply failing.

The benchmarks, in plain terms

Anthropic reports that Opus 5 roughly doubles the score of Opus 4.8 on its Frontier-Bench reasoning test at a lower cost, and lands within half a percent of Fable 5 on the CursorBench coding test while costing about half as much per task. On agent style tasks it claims a pass rate around one and a half times the next best model for the same spend.

The pattern across those numbers is consistent. Opus 5 is not trying to beat the very top of the range on raw capability. It is trying to deliver most of that capability for a fraction of the price, which is exactly the trade most teams actually care about.

Half the price of Fable 5

Fable 5 remains the more capable model on the hardest, longest problems, and it costs $10 per million input tokens and $50 per million output. Opus 5 sits at half that on both, so a workload that was borderline on cost with Fable 5 can look very different on Opus 5.

The effort setting matters here too. Running a routine job at low effort keeps the bill down, while reserving high effort for the genuinely hard work means you are not paying frontier prices for tasks that never needed them. For a sense of how these models compare on business workloads, see our breakdown of Claude versus ChatGPT.

Half the price of Fable 5

Why Australian businesses should care

For most Australian operators the interesting question is not whether Opus 5 tops a leaderboard. It is whether the same automation now costs less to run at scale. Matching invoices, drafting documents, answering routine customer questions and summarising long reports are all token heavy jobs where a halving of price changes what is worth automating.

That is where a cheaper default model earns its keep. Work that did not quite justify the spend last month can become viable this month. Teams building out finance and accounting automation in particular tend to run high volumes of repetitive tasks, and those are precisely the workloads where a lower per token price compounds.

What to watch next

The move fits a clear pattern in the market, first reported by CNBC: the frontier keeps advancing, but the bigger commercial story is capable models getting cheaper. That is what makes broad automation affordable rather than experimental.

For teams already using Claude, the practical next step is small. Check which model your plan now defaults to, test a couple of your real workflows on Opus 5 at different effort settings, and see where the numbers land. The capability was already useful. The point of this release is that it now costs less to put to work.

Scroll to Top