On this page
In 2026, a first version typically costs between $3,000 and $8,000 with a freelancer, $10,000 to $20,000 with a development shop, and $15,000 or more with a no-code agency. AI assisted studios have pushed the floor lower, under $2,500 in some cases. The larger cost is almost never the invoice. It is building the wrong thing and paying twice.
Every founder asks this question first, and almost every answer online is useless, because it quotes a number without saying what the number buys.
So here are the ranges as they stand in 2026, what sits behind each one, and the costs that never appear on a quote. The figures below are typical market ranges for a genuine first version, not a prototype and not a design file.
The short answer
A first version costs somewhere between a few thousand dollars and twenty thousand, depending less on how complicated your product is and more on who you hire to build it.
That spread is wide because "MVP" means completely different things to different vendors. Some deliver a clickable design. Some deliver a working product with authentication, a database, and a domain. The word does not tell you which, so the price does not either. If you are unclear on what should even count as version one, the guides cover that before you start ringing round for quotes.
What each route costs in 2026
| Route | Typical cost | Typical timeline | What you actually get |
|---|---|---|---|
| Doing it yourself | Your time | 2 to 6 months | Whatever you can learn to build, on evenings and weekends. |
| Freelancer | $3,000 to $8,000 | 4 to 10 weeks | Execution of your brief. Quality varies enormously by who you find. |
| Development shop | $10,000 to $20,000 | 8 to 16 weeks | A team, a process, and a contract. They build what you specify. |
| No-code agency | $15,000 and up | 4 to 8 weeks | Faster delivery on Bubble or similar. Same reliance on your brief. |
| AI assisted studio | Under $5,000 | 2 to 4 weeks | A live product, built with AI tooling. Scoping quality is the variable. |
The bottom row is the one that changed. Tools like Lovable, Cursor, and Supabase collapsed the build time for a standard web application from months into days, and the price followed. Building is no longer the expensive part.
Where the money actually goes
Founders assume they are paying for code. Mostly they are not. On a typical eight to sixteen week engagement, the build itself is a minority of the invoice.
- Coordination. Standups, handovers, project management, and the overhead of several people staying aligned on one product.
- Rework. Features built, reviewed, misunderstood, and rebuilt. This is the single largest hidden line item on most projects.
- Scope that grew. The feature added in week six because it seemed small, which quietly moved the deadline by two weeks.
- Actual building. Often less than half the total, and falling every year as tooling improves.
Which is why comparing hourly rates between vendors tells you almost nothing. A cheaper rate attached to more rework is more expensive.
The costs nobody puts in the quote
Three more show up after you sign, and they are worth budgeting for.
Your own time
A twelve week build is twelve weeks of calls, reviews, and decisions. Founders routinely underestimate this, then find themselves the bottleneck in their own project.
Running costs
Hosting, a database, a domain, email, and any AI usage. Small at first, typically twenty to a hundred dollars a month for an early product, but ongoing and in your name.
The second build
The one that happens when the first version misses. This is the cost that dwarfs everything above, and it is the only one that is genuinely avoidable.
The expensive part was never the build. It was building the wrong thing quickly.
Why the cheapest quote rarely wins
If two vendors quote $4,000 and $12,000, the instinct is to read the gap as profit margin. Sometimes it is. More often it is what happens before anyone opens an editor.
The cheaper quote usually assumes your brief is correct and prices the execution of it. The more expensive one may include discovery, scoping, and someone whose job is to tell you which half of your feature list should not exist yet.
That second thing is worth more than it sounds. A build that comes in at $4,000 and gets rebuilt six months later cost $8,000 and half a year. A build that came in higher but shipped something users actually wanted cost exactly what it said on the invoice.
Do not choose on price alone. Ask each vendor what they would remove from your feature list and why. The ones who cannot answer are pricing execution. The ones who can are pricing judgment, and judgment is the part that determines whether the money was worth spending.
What to do with this
Before you collect quotes, get clear on what version one has to do. A vague brief priced by four vendors produces four incomparable numbers, and you will end up choosing on gut feel.
- Write down the single job a user must be able to complete. That is your product.
- List everything else separately. That list is version two, and it can wait.
- Ask every vendor to quote against the first list only.
- Ask each of them what they would cut. Compare the answers, not just the numbers.
If you would rather not do that alone, it is the first week of every Firstspec engagement, and our own pricing is on the pricing page.
Figures are typical market ranges observed across studios, agencies, and freelancers in 2026, not quotes from named companies. Your own costs will vary with complexity, location, and scope. Last updated 26 July 2026.