Anthropic just dropped its latest frontier model

Anthropic officially launched Claude Opus 5, the newest addition to its Opus line, aimed squarely at coding, research, and complex business workflows. The company is making a bold claim: it rivals the much-hyped Claude Fable 5 on several key tests — while costing significantly less per task.

Opus 5 is rolling out across Claude’s apps and API. It’s now the default model for Claude Max subscribers and the strongest option bundled with Claude Pro. For users who’ve been watching Fable 5 move behind pay-as-you-go credits starting July 20, this could be a welcome shift.

How close is Claude Opus 5 to Fable 5, really?

Anthropic’s own numbers paint an interesting picture. On CursorBench 3.2 — a benchmark that tests coding agents on real software development tasks — Opus 5 finished within 0.5% of Fable 5’s best score at maximum effort. The catch? It hit that result at half the cost per task.

On OSWorld 2.0, which measures how well AI agents operate computers and complete tasks across apps and files, Opus 5 actually beat Fable 5’s best result at just over a third of the cost. That’s not a small margin.

Still, these are Anthropic’s own tests. Real-world performance can vary wildly depending on your workload, your codebase, and how you prompt the model. Treat the numbers as directional, not gospel.

What the benchmarks actually measure

• CursorBench 3.2: Real software engineering tasks, scored on completion and correctness.
• OSWorld 2.0: AI agents navigating a computer, using apps, and managing files like a human would.
• Frontier-Bench v0.1: Difficult software engineering problems, testing persistence and debugging ability.

Big leap over Opus 4.8

Compared to its predecessor, Opus 5 is a substantial upgrade. Anthropic says it more than doubled Opus 4.8’s score on Frontier-Bench, which tests AI agents on challenging software engineering tasks. It also completed those tasks at a lower average cost.

The company highlights that Opus 5 is better at checking its own work, diagnosing root causes of bugs, and pushing through difficult jobs instead of stopping after a quick fix. In one example, it caught an edge case that an existing community patch had missed. That kind of thoroughness matters when you’re debugging a production system at 2 a.m.

Pricing stays the same

Here’s the part that might surprise you: Claude Opus 5 costs $5 per million input tokens and $25 per million output tokens — unchanged from Opus 4.8. You’re getting a more capable model for the same price. For developers and teams watching API bills, that’s a meaningful win.

For Pro and standard Team subscribers, Opus 5 could fill the gap left by Fable 5 moving behind pay-as-you-go credits. If you’ve been relying on Fable for heavy lifting but don’t want to pay per task, Opus 5 is now your best in-subscription option.

Should you switch to Claude Opus 5?

If you’re a Claude Max or Pro user, you don’t need to do anything — Opus 5 is already there. For API users, it’s worth running your own evals on your specific tasks. The benchmark results are promising, but nothing beats testing on your own code.

Anthropic’s positioning is clear: Opus 5 is the cost-effective workhorse, while Fable 5 remains the premium option for the absolute hardest problems. If your budget is tight and your tasks are complex, Opus 5 deserves a serious look.

Just remember — these are vendor-reported numbers. Independent verification will come as developers push the model in the wild. Until then, consider this a strong contender, not a certified champion.

British Gas is betting big on AI — and 1,300 call centre workers are paying the price

Centrica, the parent company of British Gas, has confirmed it will cut 1,300 call centre jobs over the next two years. The company says the move is part of a “targeted deployment of AI tools” and a shift in how customers interact with the brand.

Here’s the breakdown: 800 roles will go due to AI-driven automation, on top of 500 cuts announced just last month. The job losses will hit customer service teams in Glasgow, Edinburgh, Cardiff, Leicester, Stockport, and Leeds.

Some positions will simply disappear through natural attrition — employees leave, nobody gets hired to replace them. The rest will come through compulsory redundancies. Trade unions are furious, warning that Centrica is handing hundreds of human jobs to chatbots.

What Centrica’s CEO says about AI and customer preference

Centrica CEO Chris O’Shea insists AI isn’t the direct cause of the cuts. Instead, he points to customer behaviour. According to him, more than 90% of British Gas customers now start with digital channels, and call volumes have fallen by 20% (via The Guardian).

But here’s the thing: choosing to open an app or check a website first doesn’t mean customers prefer talking to a chatbot. It just means they prefer convenience. Those are two very different things.

What the surveys actually say about AI support

The evidence that people prefer human support is pretty overwhelming. A February 2026 SurveyMonkey report found that 79% of Americans strongly preferred dealing with a person. Only 8% actively preferred AI. And 81% believed companies were introducing AI mainly to save money — not to improve service.

Another survey from Pegasystems, conducted by YouGov among more than 4,700 consumers in North America and the UK, found that 66% preferred human-led support. Around 46% said AI rarely or never solved their problem. Just 2% wanted to deal only with chatbots.

Neither survey looked specifically at British Gas customers. But both suggest Centrica may be confusing a preference for digital tools with a preference for AI. AI customer support might be efficient, but it’s not what most people actually want.

When does AI customer service actually work well?

To be fair, AI has its place. It works brilliantly when the problem is simple and the customer wants a quick answer. Research published last year cited an Intercom survey where 61% of consumers preferred AI for fast replies to routine questions.

There’s also a strong financial incentive. One 2026 industry report estimates that AI can cut the average cost of a customer interaction from $4.60 to $1.45. AI agents can also solve between 76% and 92% of basic ecommerce requests.

So yes, for checking a bill, changing an appointment, or tracking a delivery — a chatbot is often fine.

The real problem: complexity

But a serious billing dispute? An error that’s cost you money? A vulnerable customer who needs extra help? That’s where AI falls apart. The same surveys that show preference for humans also show that satisfaction plummets when a chatbot fails to resolve a complex issue.

British Gas customers may well be happy to check a bill online. But when the problem becomes complicated, most people still want to speak to a human being. The available evidence is pretty clear on that.

Centrica’s move might save money in the short term. But if customers feel forced into chatbot vs human support situations they don’t want, the long-term cost could be far higher.

Trade unions have called for more transparency around how AI decisions are made. They’ve also asked Centrica to guarantee that vulnerable customers can still reach a human agent. So far, the company hasn’t given clear answers.

What this means for the future of customer service

This isn’t just about British Gas. It’s a signal of where the industry is heading. Call centre AI trends show that automation is being rolled out across sectors — telecom, banking, retail, utilities. And in every case, the same tension appears: companies want to cut costs, customers want to talk to people.

The smart approach probably isn’t all-or-nothing. It’s about routing simple queries to AI and escalating complex ones to humans quickly. But that requires investment in intelligent routing, not just cutting staff.

For now, British Gas customers might want to keep the phone number handy. The chatbot might be able to handle a meter reading. But when something actually goes wrong, you’ll likely want a human — and there will be fewer of them available.

The Scoop: A Speaker With a Personality

OpenAI’s long-rumored jump into physical products finally has a shape — and it’s not a phone. According to a Bloomberg report, the company’s first OpenAI hardware device is a mobile smart speaker with no screen. It’s being pitched internally as a “humanlike AI companion that lives in the home.”

The device is still under development, but sources describe it as having a “personality.” It can proactively learn about its owner over time, pulling from emails and other digital data to deliver increasingly personalized responses. Think of it as ChatGPT given a body — and a bit of autonomy.

What Makes This Speaker Different?

Most smart speakers sit on a shelf and wait for a wake word. This one, reportedly, does more. Bloomberg’s sources say the device includes “mechanical elements that can move on their own.” That’s a deliberate design choice, meant to make the gadget feel less like a tool and more like a presence — a “physical manifestation of ChatGPT,” as one insider put it.

It’s a strange concept, sure. But it also signals where OpenAI thinks consumer AI is headed: not just answers, but companionship.

The Apple DNA Inside

Notably, the project was built with help from former Apple engineers — the kind who worked on the iPhone and the Mac. That pedigree explains the polish OpenAI is aiming for. But it also makes the timing awkward.

The Elephant in the Room: Apple’s Lawsuit

Last week, Apple sued OpenAI, accusing the AI company of stealing trade secrets. Apple’s legal team called the initial allegations just “the tip of the iceberg,” hinting that more would surface during discovery. OpenAI has denied any wrongdoing.

So how does the new device fit into that mess? Bloomberg’s sources say OpenAI believes its product “veers significantly from anything Apple has on the market today” and that it’s “unlikely” to violate Apple’s trade secrets. Whether a judge agrees is another matter entirely.

This isn’t just a legal headache — it’s a strategic one. OpenAI has long hinted at building its own hardware, with some rumors pointing to a phone that would go head-to-head with Apple. The speaker is a less confrontational move, but the lawsuit still casts a long shadow over the launch.

The Bigger Picture: Money Is Flooding Into AI Hardware

OpenAI isn’t the only one chasing this vision. The broader tech world is suddenly obsessed with consumer AI hardware. Consider Hark, an AI lab founded by Brett Adcock. In May, it raised an oversubscribed $700 million Series A at a $6 billion valuation. The goal? Build “personal intelligence” — proprietary AI models paired with custom hardware, framed as a “universal interface between humans and machines.”

Hark hasn’t even revealed what its device looks like yet. That’s how much capital is chasing this category before a single product ships.

What This Means for the Smart Speaker Market

If OpenAI pulls this off, it could shake up a market dominated by Amazon Echo and Google Nest devices. Those products are useful, but they’re not exactly companions. A screenless speaker that moves and learns could carve out a new niche — if the price is right and the privacy concerns don’t sink it.

Questions Still Up in the Air

Plenty of details remain unknown. When will it launch? What will it cost? How will OpenAI handle the data it collects from your emails and daily life? And will the Apple lawsuit delay things?

For now, we know this: OpenAI’s first hardware device isn’t a phone, and it isn’t a boring box. It’s a speaker with a soul — or at least, a very convincing simulation of one.

Keep an eye on this space. The home AI race is just getting started.