Lovable's shift towards value-based pricing

Welcome back to Good Better Best.

Each week, we break down real pricing, packaging, and product moves from SaaS and AI leaders and share the ideas worth stealing.

This week, several people sent me Tomasz Tunguz’s latest post about selling inference, where he rekindled one of the oldest arguments in pricing for the age of AI: cost-based vs value-based pricing. This hasn’t really been a debate for a while (value was the obvious choice in legacy SaaS), but the nature of inference makes it trickier.

On one hand, model costs are being commoditized, and a value-based approach avoids a race to the bottom. On the other, model costs are public and competition is ruthless, so there’s a case to be made for being cost-based and giving buyers transparency.

In practice, most companies don’t sit at either extreme, and the interesting decisions happen in the space between. Lovable made two pricing changes this week that show what moving through that middle looks like, and we break them down below.

Let’s get to it.

PS. Ulrik and I recorded a new episode of Pricing Page unPacked, breaking down Intercom’s Fin. Note: the episode was recorded before the Salesforce acquisition. Listen wherever you get your podcasts.

This week we observed 100+ changes. The highlights:

Check out more updates on PricingSaaS →

Here’s what was top of mind in Pulse this week:

In his piece, Tunguz makes a strong case for value-based pricing. His argument is that once the raw token is a commodity, which it increasingly is, a markup on top is fragile. The customer can look up the API price and route around you, and as inference keeps getting cheaper, the markup compresses toward nothing.

The durable alternative is to stop pricing on the token at all. Charge for the work instead — the outcome the customer actually wants. Outcome-based pricing is the ideal (e.g., Sierra charges per resolved ticket, Fin charges $0.99 a resolution). The customer buys a result and never sees the inference bill underneath it.

But for a variety of reasons, outcome-based pricing isn’t realistic for most companies right now. The real question is how far in that direction you can actually move, while meeting customers where they are, and being realistic within your market.

Lovable is an AI app builder (an increasingly competitive space) and heavy reseller of inference. Their credit model has become a common reference for AI pricing, and they iterate regularly. This week, they made two changes that push from cost-based towards value-based pricing.

Lovable used to bill on two tracks: a monthly subscription for build credits, plus a separate, usage-based dollar bill for the Cloud hosting and runtime AI the deployed app consumes. Now there's a single credit balance. Building, hosting, and runtime AI all draw from the same pool, with grants (daily build credits, monthly cloud grants) covering light usage before tapping into the plan balance.

The move shifts the conversation for customers from buying tokens and compute hours to buying capacity to build and run apps. It's the same idea Snowflake and Databricks have used with credits for years: put a clean, durable unit between your price and your raw cost.

Lovable also formally differentiated between credits on the Pro plan (Pro Credits) and credits on the Business plan (Business Credits). While this was inferred previously, the update makes it clear that these credits should not be valued the same. Pro credits are worth $0.25 each, and Business credits are worth $0.50 each (top ups are $0.30 and $0.60 respectively).

Importantly, the underlying Cloud and AI cost the same on both plans. A Business credit just costs twice as much. This premium is how Lovable charges for the Business feature set: SSO, role-based access, a security center, governance. It ties what customers pay to the value of the platform, not to the cost of the inference underneath.

Both moves shift Lovable’s model toward value, while giving customers the transparency they need, and being realistic about how far they can push towards outcomes. Plan-specific credits clarify the premium for business features (SSO, governance, security), but stops well short of outcome-based pricing. Lovable isn't charging per app shipped or per workflow automated; it's still selling credits, just pricing the premium tiers higher.

The unified credit abstracts the volatile usage bill into one clean unit, while giving devs actionable visibility into credit burn by category.

Put them together and you get a company iterating toward value on both fronts. Cost versus value is rarely a binary you pick once, but a dial you adjust over time, and the right answer likely lies somewhere in the middle.

Thanks for reading! If you’re working on AI monetization and want to learn more about how we help, book time here.

Until next time,

Rob