Saw the new pricing announcement this morning. I'm trying to figure out what this means for conversations I'm already having with a few enterprise clients in the region.
Context: I'm building AI features into our SaaS platform for customers across Southeast Asia. Most of them care deeply about data residency, compliance, and predictable costs. For the last few months, I've been running cost models and demo workflows using Claude through the API, and it's been solid for the use case. Now with the pricing shift, I need to recalculate everything.
Here's what's frustrating me. It's not that the pricing changed. Pricing changes. It's that I'm in the middle of client conversations where I've already mentioned cost expectations based on the old structure. One customer specifically asked about per-token economics last month, and I gave them numbers. Now I need to either go back and revise those numbers or decide whether the new pricing still makes sense for what they're trying to do.
The other issue is more subtle. Several of my clients are based in countries with strict data residency requirements. They want everything to stay within their borders. Using Claude API means their data goes through Anthropic's infrastructure, which I can explain and defend, but it adds friction to conversations. If I'm going to ask them to make that trade-off (residency concerns vs. better model performance), the cost needs to be justified. When the math changes, the trade-off becomes harder to justify.
I've been using Claude 4.7 and Sonnet 4.6 pretty heavily. Both perform well for what I'm doing. I don't have strong feelings about staying married to Anthropic's models, but I also don't want to switch horses mid-stream if I don't have to. The question is whether the new pricing keeps them competitive against, let's say, OpenAI or even fine-tuned open-source approaches.
Two specific questions for anyone else working with enterprise clients in similar situations:
One, have you already had to revisit cost conversations with clients because of this change? How did that go? Did clients seem to expect this kind of volatility, or did it come as a surprise?
Two, if you're choosing between closed API providers for enterprise work where cost predictability matters, what's your actual decision framework? Is it model quality, cost, data handling, compliance features, or some weighted combination?
I'm not panicking. The pricing is still reasonable, and Claude's quality is still strong. But there's a coordination problem here that bugs me. I can't finalize contracts or commit to technology choices until I know the cost baseline, and that baseline keeps shifting. My enterprise customers don't move fast anyway. They want stability and clear numbers, and right now I can't give them the second one.
Maybe this settles down in a few weeks and I'm overthinking it. But I'd like to hear from others about how they're thinking through this.