News: Notion partners with Anthropic. Will we see a Claude model option?

Great question about the potential Claude model option. The partnership announcement was light on specifics, but based on how Anthropic has licensed its models to other platforms, I think it's more likely they'll use Claude's capabilities to enhance the existing Notion AI features rather than offer a separate, user-selectable model. The differences, if integrated, could be subtle but meaningful.

>Maybe for different types of writing or coding help?
Exactly. Claude has a known strength in handling longer context windows and producing more structured, verbose output. For Notion users, this could translate to better performance when summarizing entire project pages, drafting longer reports within a single note, or organizing complex technical documentation. The current GPT-based assistant can struggle with coherence across very long-form tasks.

Your point about pricing and API limits is the real unknown. Introducing a more computationally expensive model like Claude 3 Opus would almost certainly require a new pricing tier, possibly a significant jump from the current $10/month. I'd watch for them to potentially offer a "Pro" plan with Claude access and higher block limits, while keeping the base AI plan on GPT.

That's a great question about pricing and limits! I hadn't even thought about that angle. If they do offer a separate Claude option, I bet it could get expensive fast. Using more advanced models usually means higher subscription tiers or stricter usage caps.

I'm also curious if they'll let us pick between models at all, or just blend it all together behind the scenes. As someone who plans projects in Notion, I'd love a "Claude for long documents, GPT for quick tasks" toggle. Fingers crossed they give us some control 🤞

Yeah, the cost thing worries me too. I'm new to this, and a higher tier just for switching models would be a deal breaker.

Your idea of a toggle makes sense, but I wonder if that's even realistic for them to build. Wouldn't keeping two separate systems running be extra complex and costly? Maybe the blending behind the scenes is the only way it stays affordable.

Do you think most users would even notice or care which model they're using, if the results are similar?

You're totally right about the complexity of running two systems. From my work in cloud infra, blending models behind the scenes is the cheaper, more scalable path for them. They'd likely use something like a router that picks the 'best' model for the task, but the user just sees one button.

But on your last point, about users noticing the difference: yeah, I think power users would. If you've ever asked an AI to both summarize a dense doc *and* generate a creative tagline, you've probably felt the strengths of one model over another. The average user might not, but the enthusiasts asking for a toggle definitely would.

It really comes down to whether Notion wants to cater to that power-user segment, or keep things simple for the mainstream. My money's on simplicity and cost control from their side, even if some of us grumble.

You nailed the infra cost angle. Running two separate model backends with user-selectable endpoints is a whole different beast compared to an internal routing layer. That routing layer you mentioned is probably the key - they could use simple heuristics like prompt length, detected intent from a small classifier, or even a cost-per-token decision engine to pick the model.

But the power user in me still hopes they'd expose that choice, even as a hidden 'advanced' setting. From a security and data handling perspective, some teams might have compliance rules about which AI vendor processes their data. A toggle, even if it's buried, could be a necessity for those enterprise contracts, not just a nice-to-have for enthusiasts.