Convex vs Supabase vs Firebase for Indie Hackers in 2026

Originally published at devtoolpicks.com

Picking a backend is one of the first real decisions you make on a new project, and it's annoyingly hard to reverse. Switch later and you're rewriting your data layer, your auth, and half your queries. So it's worth getting right the first time.

For most indie hackers in 2026, the choice comes down to three tools. Firebase is the old default from Google. Supabase is the open-source Postgres challenger that ate a huge chunk of Firebase's audience. And Convex is the newer entrant that does something neither giant does by default: real-time sync, baked in, with no WebSocket plumbing.

They look similar from the outside. Database, auth, file storage, serverless functions, all managed for you. But they're built on completely different foundations, and that difference shows up in two places that matter to a bootstrapped founder: how you model your data, and how your bill behaves when things go well. I've shipped on the Postgres side of this fence, so I'll be upfront about my bias and then give each one a fair shake. If you only want the two-way on the incumbents, I went deeper in Supabase vs Firebase. This post adds Convex and the real-time angle.

The Quick Verdict

Short version: Supabase is the safe, predictable default for most SaaS apps. Convex wins when real-time sync is the product. Firebase ships fastest but its bill is the least predictable of the three.

Convex

Convex is the interesting one because it isn't trying to be a managed database. It's a reactive backend. You write your queries and mutations as plain TypeScript functions that run inside Convex's own runtime, and any UI component that subscribes to a query updates automatically when the underlying data changes. No polling. No WebSocket server. No pub/sub setup. For a collaborative app, that's weeks of engineering you simply don't do.

It's fully typed end to end, includes file storage and vector search, and has built-in scheduling and cron for background jobs. The free tier is genuinely generous: roughly 1 million function calls a month, and you can have up to 6 people on a team before paying anything. There's also a startup program that gives qualifying teams up to a year of Pro for free.

Pricing is where you need to read carefully. The Professional plan is $25 per developer per month. That "per developer" matters. A solo founder pays $25. A three-person team pays $75. On top of that, heavy real-time workloads rack up usage costs: database bandwidth runs around $0.20 per GB and action compute around $0.30 per GB-hour, and EU regions cost 1.3x.

Who should skip it? Anyone who needs SQL. Convex uses a proprietary document model, so there's no Postgres, no joins the way you know them, no existing ORM, and no pgvector or row-level security. Migrating an existing Postgres app to Convex is a full rewrite, not a port. And because the reactive architecture is proprietary, leaving later means re-engineering your client-side sync logic. You're trading portability for a real-time superpower. If real-time isn't core to what you're building, that's a bad trade.

Supabase

Supabase is the one I reach for, and it's the one most indie hackers land on. The pitch is simple: it's a real PostgreSQL database with auth, storage, edge functions, and real-time subscriptions wired on top, all behind a clean dashboard with auto-generated APIs. Because it's actual Postgres, you get standard SQL, foreign keys, row-level security, and pgvector for AI embeddings. Nothing proprietary about the core. And it's open source, so you can self-host the whole thing on your own server if you ever want to.

The free tier is strong: 500 MB database, 50,000 monthly active users for auth, 1 GB file storage, and 2 projects. There's one catch you have to plan around. Free projects pause after 7 days of inactivity. Your data is safe and you can resume it, but anything user-facing can't live on the free plan. That's the nudge to Pro.

Pro is a flat $25 a month per organization, and that flat number is the headline. It includes $10 in compute credits that cover one Micro instance, so a single production app is effectively $25 to $35 all in. Unlike Convex, the base price doesn't climb with your team size. Where it does climb is compute per extra project and usage overages on bandwidth, so most real apps land around $35 to $75 a month once they have traffic. The Team plan jumps to $599 a month, but that's for SOC 2 and HIPAA compliance, not features you'll miss as a solo dev.

Who should skip it? If your app is fundamentally real-time and collaborative, Supabase's real-time is bolted onto Postgres rather than the default, so you'll do more setup than Convex asks for. Edge functions also have a noticeable cold start of a few hundred milliseconds after idle. And if you want raw serverless Postgres without the bundled auth and storage, a focused database like Neon or Turso may fit better. I compared those in Turso vs Neon vs Supabase, and there are more options in the best Supabase alternatives.

Firebase

Firebase is the original "backend without a backend," and it's still the fastest way to get a mobile or web MVP live. Auth, Firestore, hosting, cloud functions, analytics, and crash reporting all snap together, and the SDKs are everywhere. If your goal is to ship something this weekend and you're already in Google's world, nothing gets you to a working app faster.

The pricing model is the opposite of the other two. There are no tiers and no seats, just two plans. Spark is the free plan: 50,000 Firestore reads a day, 20,000 writes, 1 GB storage, and 50,000 monthly active users for auth, with no payment method required. Blaze is pure pay-as-you-go. It includes the same free allowances, then charges for usage above them: Firestore is $0.06 per 100,000 reads, storage runs a few cents per GB, and functions and bandwidth meter separately.

For a small app, that often means a $1 to $10 monthly bill, which is great. The problem is the shape of the model, not the rates.

Who should skip it? Anyone who needs to bound their spend. Firestore bills per operation, and Blaze has no hard spending cap. Budget alerts warn you, but they don't stop charges. Real-time listeners and AI features that re-read the same data amplify your read count fast, so a single user action can fire dozens of billable reads, and a viral day or a runaway loop can turn a $10 bill into hundreds before you notice. On top of that, Firestore's NoSQL model makes relational queries awkward, you can't self-host, and migrating off after a year of production is genuinely painful. The newer Data Connect feature adds a Postgres option, but it's still early. If you're selling a fixed-price subscription, an unbounded backend bill is the wrong foundation. There are calmer relational options in the best PlanetScale alternatives.

How Do the Three Actually Compare?

The marketing pages all promise the same thing. Here's where they genuinely differ.

Two patterns fall out of that table. On data model, this is really a SQL versus document decision. If your data is relational and you think in tables and joins, Supabase is home. If your data is document-shaped and real-time is central, Convex or Firebase fit better. On cost, predictability runs Supabase, then Convex, then Firebase. The flat $25 is the easiest number to budget against, the per-developer model is predictable but scales with hiring, and the uncapped pay-as-you-go model is the one that can ruin a month.

What Does Each One Actually Cost?

Run the numbers for a real team and the pricing models separate fast. A solo founder pays $25 on any of the three paid paths, so cost isn't the deciding factor at the start. It changes when you add people. Convex charges per developer, so a three-person team is $75 a month and a five-person team is $125, before a byte of usage. Supabase holds its flat $25 base no matter how many people touch the project, with compute added per extra deployment. Firebase has no seats at all, so team size never moves the bill, only usage does, and without a spending cap that's the number you have to babysit.

So the cost story splits by who you are. A solo founder building real-time? The $25 on Convex is a non-issue and you get sync for free. A small team that plans to hire? Supabase's flat base is the cheapest to grow into, since hiring your fourth developer doesn't touch the bill. Shipping a throwaway prototype? Firebase on Spark is genuinely free until you scale, which is exactly when you switch on budget alerts.

How Do You Choose the Right One?

Start with your data, then layer in cost sensitivity and team size.

View the interactive diagram on devtoolpicks.com

A few concrete scenarios. Building a collaborative editor, a live dashboard, or a multiplayer tool where state has to stay in sync? Convex saves you the most work, and as a solo founder the $25 is fine. Building a normal SaaS with users, billing, and relational data? Supabase, almost every time, because Postgres and a flat predictable bill are exactly what you want. Throwing together a mobile MVP this weekend to test an idea? Firebase will get you there fastest, just turn on budget alerts before you launch.

The Verdict: Which Should You Use?

For most indie hackers, Supabase is the default pick. Real Postgres, a flat $25 a month, open source so you're never trapped, and a feature set that covers the vast majority of SaaS apps. The 7-day free pause is the only sharp edge, and it disappears the moment you go to Pro. It's the safest long-term bet of the three.

Choose Convex when real-time sync is the product, not a feature. If you'd otherwise spend weeks building WebSocket infrastructure and state synchronization, Convex hands you that for free, and the TypeScript-native model is a genuine pleasure. Just go in knowing you're accepting a proprietary model and per-developer pricing as the cost of that speed.

Choose Firebase when raw speed to a first version matters more than anything else and you're comfortable in Google's ecosystem. It's the fastest on-ramp, and for a prototype that may never need to scale, the free tier is hard to argue with. The day it becomes a real business, set strict budget alerts and start watching your read counts, because the bill won't watch itself.

The honest meta-point: the best backend is the one that stays boring while your app gets interesting. For a bootstrapped founder, boring usually means predictable cost and an exit if you need one. That's why Supabase wins for most people, why Convex wins for real-time, and why Firebase asks for the most caution. Building something with one of these? Tell me which you picked over on @devtoolpicks.