Anthropic just made a bet that will matter to every team trying to put AI to work: on July 24, 2026 it launched Claude Opus 5, a model that gets close to the intelligence of its flagship Fable 5 while charging half as much to run. For anyone who has watched their token bill climb as they scale automation, that is the headline that counts.
Opus 5 costs 5 dollars per million input tokens and 25 dollars per million output tokens. Its bigger sibling, Fable 5, runs 10 and 50. On CursorBench 3.2 at maximum effort, Opus 5 lands within half a percent of Fable 5’s peak score while costing far less per task. It is the fourth model Anthropic has shipped in two months, and the company is aiming it squarely at coding and automation rather than casual chat.
What ‘proactive’ really means
Anthropic is calling Opus 5 a proactive model, and the word is doing real work. Instead of answering a prompt and stopping, Opus 5 is built to verify its own output, iterate until a task actually succeeds, and push through multi-step problems without a human nudging it at every turn. In one example Anthropic shared, the model was handed a drawing of a machine part with no direct way to view it, so it wrote its own computer vision pipeline to pull the geometry out of the raw pixels and reconstructed the full part on its own.
That behavior is the difference between a chatbot and an agent. A chatbot gives you a draft. An agent finishes the job, checks its work, and hands you a result you can use.
Where Opus 5 pulls ahead
The benchmarks Anthropic chose to highlight say a lot about who this model is for. On OSWorld 2.0, a test of using a real computer, Opus 5 beats Fable 5’s best result at roughly one-third the cost. On Zapier’s AutomationBench, its pass rate is about 1.5 times the next best model, and it hits 100 percent on account-health workflows where earlier models failed outright. On coding, it doubles the previous Opus generation’s score on Frontier-Bench.
None of those are chat benchmarks. They measure whether a model can drive software, complete a workflow end to end, and hold context across a long, messy task. That is exactly the work businesses want to hand off.
What it means for your business
Here is the pattern worth noticing: the price of frontier-level intelligence is falling by half roughly every couple of months. When the model gets cheaper and more capable at the same time, the model stops being your bottleneck. Your bottleneck becomes how quickly you can wire that intelligence into the processes that actually run your business, the reports, the analyses, the follow-ups, the workflows.
That is a strategy question, not a technology question. The teams that win the next year will not be the ones with access to the smartest model, because everyone will have that. They will be the ones who already built the plumbing to point a capable, affordable model at a real task and let it run.
For sales teams tired of cold leads, slow customer responses, and manual processes, Dapta is the ultimate tool.
Dapta is the leading platform for creating AI sales agents specifically designed to increase inbound lead conversion. Respond to your leads in less than a minute with voice AI and WhatsApp that converts.
If you want your team to sell more while AI handles the complex stuff, you have to try it.
For a growing company, a model like Opus 5 is an invitation to revisit every place you assumed automation was too expensive. Lead qualification, customer follow-up, internal reporting, data cleanup: the workflows that were borderline at last quarter’s prices are comfortably in reach now. The question is no longer whether the AI is good enough or cheap enough. It is whether you have a way to deploy it.
Anthropic is clearly racing to own the automation layer, and cutting the price of its near-flagship model in half is how it plans to get there. For teams in LATAM and global markets watching costs closely, cheaper frontier intelligence is not an abstract milestone. It is a reason to move.
