AI buildersOct 26, 20263 min read

Claude Code pricing: what it costs to keep building after Lovable

Claude Code has no free plan. The cheapest way in is Claude Pro at $20 a month. Above that there is Max at $100 or $200 a month, and there is the API, where you pay for every token. If you are coming from Lovable, that is only part of the bill: Claude Code edits your code, but it does not host the app or run a database.

What Claude Code costs after Lovable

The plans

Plan	Price	Claude Code	Usage
Free	$0	No	Chat only
Pro	$20 a month, or $200 a year ($17 a month)	Yes	At least 5x Free per five-hour session
Max 5x	$100 a month	Yes	5x Pro per five-hour session
Max 20x	$200 a month	Yes	20x Pro per five-hour session
Team, Standard seat	$25 per person a month, or $20 billed yearly. At least 2 people	Yes	1.25x Pro per five-hour session
Team, Premium seat	$125 per person a month, or $100 billed yearly	Yes	6.25x Pro per five-hour session
Enterprise	$20 per seat a month plus usage at API rates, billed yearly. At least 20 seats	Yes	Billed by usage
API (Claude Console)	Pay per token	Yes	No plan limits, you pay for what you use

Prices checked on 26 October 2026, in US dollars, without tax. Max is monthly only. Anthropic changes plans often, so check the pricing page before you pay.

Pro includes Opus, Sonnet and Haiku. Fable, the biggest model, is not part of the Pro limits: on Pro it bills to usage credits. On Max, Fable has its own weekly limit, which the pricing page puts at 50% of the weekly limits.

How the limits work

Pro and Max have a five-hour session limit and a weekly limit. Team seats have the same two. There is no fixed number of messages. Claude Code sends the whole conversation with every request, so a long session in a big project uses the limit much faster than a few small fixes.

The part people miss: the limits are shared with the Claude chat apps. If you spend the morning chatting on claude.ai, you have less left for coding in the afternoon.

The /status command in Claude Code shows what is left. On Pro and Max, /usage shows your plan usage and what used it (skills, subagents, plugins and MCP servers), with suggestions.

When you hit a limit, you can:

Wait for the reset.
Move to a bigger plan: Pro to Max 5x, or Max 5x to Max 20x.
Turn on usage credits and keep working at standard API rates. Run /usage-credits in Claude Code, and on Pro or Max it opens Settings, Usage on claude.ai, where you can also set a monthly spend limit.
Switch to a Claude Console account and pay for API usage.

Paying per token through the API

If you log in to Claude Code with a Claude Console account instead of a subscription, you pay API prices. There is no five-hour or weekly limit. You pay from prepaid credits, and when they run out the API stops working until you add more. Auto-reload buys more on its own when the balance gets low, so check whether it is on in Console Billing.

API prices per million tokens:

Model	Input	Output	Cache hits
Fable 5.1	$10	$50	$0.25
Opus 5.5	$4	$20	$0.20
Sonnet 5.5	$2	$10	$0.10
Haiku 5.5 (prompts up to 100,000 tokens)	$0.10	$0.50	$0.01

The cache price matters because Claude Code caches the repeated parts of each request on its own. The conversation it sends again with every message is mostly billed at the cache rate.

Anthropic publishes one average: across company deployments, Claude Code costs about $13 per developer per active day and $150 to $250 per developer a month, and stays under $30 per active day for 90% of users. That is people who code every day. If you fix things in your app on weekends, you will be far below it.

One more thing about older price comparisons: Claude 4.7 and newer models use a tokenizer that makes about 30% more tokens from the same text. So the price per token does not tell you the whole bill when you compare with older models.

If the ANTHROPIC_API_KEY environment variable is set on your computer, Claude Code asks whether to use that key. If you say yes, you pay API rates instead of using your subscription, and Claude Code remembers the choice. If your Lovable app calls Claude, keep that key in the app secrets, not in your shell.

Max also includes $100 or $200 a month in API credits. Those credits cannot be used in Claude Code or the Claude apps, not even for usage credits. They are only for building on the Claude API.

What changes in your bill when you leave Lovable

Lovable puts everything in one credit balance: the AI that builds, hosting and the built-in backend (Lovable Cloud). Its Pro plan starts at $25 a month for 100 credits. When you move to Claude Code, that one bill splits into three.

The AI: a Claude plan, from $20 a month. This replaces the credits you spent on prompts.
Hosting: Claude Code does not host anything. You pick a host, for example Vercel, Cloudflare Pages or your own server. Where to host a Lovable app compares them.
The database: if your app uses your own Supabase project, nothing changes. If it uses Lovable Cloud, the database and logins keep running on Lovable credits as long as they stay there. To stop paying Lovable for them, follow Move Lovable Cloud to your own Supabase project.

You do not have to cancel Lovable on day one. Lovable and Claude Code can work on the same GitHub repository, and editing the code yourself costs no Lovable credits. Out of Lovable credits? explains how that works.

Which plan to pick

For someone who leaves Lovable with one or two apps:

Start with Pro, monthly. $20 is less than the smallest Lovable Pro plan, and in a week you know if the limits are enough. Do not pay for the year before you know how you work.
Move to Max 5x if you hit the five-hour limit every day. Not once a week. Every day. Waiting for a reset now and then is cheaper than paying five times more.
Use the API if you code in bursts or run Claude Code in scripts. A few heavy days a month can cost less than Max. Keep auto-reload off until you know what a normal day costs you.
Max 20x is for people who work in Claude Code all day.

How to make your plan last longer

These come from Anthropic's own docs:

Run /clear when you switch to an unrelated task. Old context costs tokens on every next message, and /clear itself costs nothing.
Use /compact to shrink a long conversation. You can tell it what to keep, as in the example below.
Switch to Sonnet for normal work with /model. Opus is the default, but Anthropic says Sonnet handles most coding tasks well and costs less.
Check /usage. On Pro and Max it shows what used your limits and suggests changes.
Keep CLAUDE.md short. Claude Code loads it at the start of every session, so it is part of every request. Anthropic suggests under 200 lines. Take your Lovable Knowledge to CLAUDE.md or AGENTS.md shows what to bring over from Lovable.

/compact Focus on code samples and API usage

To stay strictly inside your subscription, log in with your Pro or Max account only and do not add Console credentials. Anthropic says you then are not offered API credits when you reach a limit. If you were logged in with a Console account before, run /login in Claude Code and pick your subscription.

Is it worth it

For $20 a month you get an AI that works on the real code, in a folder you own, with no credit counter on every prompt. You also take on hosting and the database. For most small apps that is a fair trade. If you never want to open a terminal or think about where the app runs, stay on Lovable and buy credits.

To compare tools first, read Should you leave Lovable for Claude Code?, Codex vs Claude Code and Cursor vs Claude Code. When you are ready, the step-by-step guide takes you from the GitHub export to hosting and the database in 6 to 19 steps.