Pricing pages in the AI-IDE category have settled into a pattern. Free tier offers an old model. Mid tier unlocks the current flagship. Top tier unlocks the reasoning models. The implied promise: pay more, get a smarter AI.
Pearl doesn't do that. Every paid plan — Pro, Business, Enterprise — gets access to every premium model we support. Claude 4 Opus, GPT-class frontier, Gemini Pro, the lot. The differentiation between tiers is quota, team features, and SLAs.
The internal decision
We had the conversation. "Should we save model X for the Business tier? It'd push more upgrades." The reasons we said no:
- It rewards spend, not work. A solo developer on Pro can need the flagship model for the same reasons a Business customer can. The work doesn't change because the bill does.
- It makes the product worse for everyone. Forcing a Pro user onto a weaker model means the answer they get is worse than the one they're paying for. Every "I downgraded my model because of pricing" is a story we don't want.
- It creates a perverse incentive on the roadmap. If "next-gen flagship" is a Business-tier upsell, we're motivated to ship it to Business first and Pro later. That's a Cursor-ism we're explicitly not copying.
- The cost difference is real but bounded. A flagship model costs more per token than a fast model. But the Pro plan's quota is sized so the math works at the flagship rate too. We sized quota down — we did not gate models.
The customer pitch
The way we say it on the pricing page:
"Differentiation is quota, team features, and SLAs — not model lock-out."
The way we say it to ourselves in product reviews:
"If a feature would make the answer to a developer's question worse, it's not a feature — it's a tax."
What this means for the roadmap
When a new flagship model lands at Anthropic or OpenAI, Pearl exposes it to every paid plan on the same day. If we can't afford to do that at our current quota math, we adjust the quota — not the access. The model picker doesn't show "available on Business+" badges. The model picker shows the models you can use.
The honest caveat
This isn't free. There's a reason the AI-IDE category settled into the gated-model pattern: it works as a pricing lever. We're betting that not doing it is a stronger differentiator than the revenue it would have produced — that the developers we want as long-term customers are the ones who'd be insulted by being asked to pay more for a smarter answer.
If that bet is wrong, we'll find out. But it's the bet we made.
Try it: open the model picker on the Pro plan (or the trial install). The full premium lineup is right there. Pick the one that fits the task — not the one your tier permits.