Video of the Week: Anthropic's Claude Opus 4.6

Welcome to a new frontier in the battle between the frontier labs - beef.

Specifically, 'beef' between Anthropic (which launched its very, very good Claude Opus 4.6 model this week) and OpenAI (which launched its very, very good Codex 5.3 model this week). As the tech gets better, the stakes get higher - these things cost billions to create and are quickly superseded by other models so the lifespan of that specific investment and launch can be days rather than months of direct RoI. It was only just before Xmas that Google landed Gemini 3 which was jaw-droppingly good - now Claude is making all the running with its associated Cowork tool that can go and do things for you, autonomously, for hours.

Which brings us back to the beef. One thing Claude does not do is voice (yet). So when it dramatises a chat interaction, as if between two people (then launches these as ads at the Super Bowl), the slight delay in the responses of the second 'person' indicates that this must be ChatGPT. Although this is never made explicit, the fact that they then start talking about products that might be useful to the person asking the question doubles-down on the shade thrown by Claude at ChatGPT.

The context: Sam Altman, OpenAI's CEO, has said that they are going to introduce ads to ChatGPT. He says that these will be clearly labelled and will not spill into the actual answers that are given by the model - and that this is merely a way to enable access for those that cannot afford to pay the subscriptions each month to access the amazing technology. Seeing ads in a freemium product is not new - Google has built a trillion-dollar company on this idea, for example, as has Facebook - but these platforms don't (didn't) have AI giving direct answers that might be skewed by a commercial relationship.

It's a balancing act and one that - candidly - we face here as we build out MyPorta. We have been really clear - no agent can pay us money for their properties to appear at the top of search results. This is non-negotiable and is the absolute bedrock of what we're doing. In essence, trust is the product - if consumers suspect that any commercial relationship is corrupting the search results (such as surreptitious 'featured properties' or over-promotion of corporate agencies as recommended to sell their home) then the hit to trust will be swift and deadly. Consumers just won't use the site.

From this core principle, a number of things also become true:

• MyPorta has to always be free for agents to list; if we allow just one agent to pay then we're inviting corruption of search results and an exodus of traffic

• The pay-to-list traditional portals are in trouble; consumers will, in the end, vote with their attention. Why would they want to see a subset of properties that are organised in order of which are most profitable to the portal, rather than the ones that are most relevant to the searcher?

Completely separate marketplaces of products and services are straightforward to signpost and organise but they're never near the purest property search you'll find. And this is perhaps where this battle between the two tech giants will inevitably end up - with both claiming they have stuck to their guns whilst enabling them to generate the money to continue furthering their ambition.

Meantime this ad, the creepiest of the four released by Anthropic, will probably live in your head as the absolute reason why paid-for, featured ads of any description can't be part of the future of any aggregator trying to operate in the new world of AI ...

🤮

---

Join Mal on the 'AI in EA Tour' to hear - live - how you can use AI tools right now to increase market share, cut costs and improve revenue per transaction. Register for a location near you here: www.myporta.ai/tour26

---