LlamaParse pricing: credits per page, and the real cost of a parsed CV

Published 21 September 2026

LlamaParse pricing, as of September 2026, is a credit system: 1,000 credits cost $1.25, and every page you parse spends credits at a rate set by the tier you chose — 1 credit a page on Fast, 3 on Cost-effective, 10 on Agentic and 45 on Agentic Plus. On top of that sits a plan: Free at $0 with 10,000 credits a month, Starter at $50/month with 40,000, Pro at $500/month with 400,000, and a custom Enterprise tier, with pay-as-you-go credits above the included block on the paid plans. So the published answer to "what does LlamaParse cost" is a rate card, not a per-document price, and the per-document price depends entirely on which tier you run and what you are trying to get out of the page.

That last part is where most cost estimates for CV parsing go wrong, and it is what the rest of this article is about. Parsing a CV to markdown is cheap on LlamaParse. Getting typed resume fields out of it — employer, title, start month, whether the role is current — is a second, separately priced step, and once you price that step the numbers look quite different.

We build ResumeJSON, a per-parse resume parsing API, so read this as an interested party's write-up. Every LlamaParse figure below was read off LlamaIndex's own pricing and documentation pages in September 2026; their pages are theirs to change, so re-check anything you are about to commit a budget to.

What LlamaParse pricing looks like on the page

The conversion rate is stated plainly in the docs — "$1.25 per 1,000 credits" — and the plan page describes the model in one line: "The LlamaIndex platform operates on a credit-based system, where each action—such as parsing, indexing, or extracting—costs a certain number of credits."

Two details in that table matter more than the headline prices. The Free plan's 10,000 credits a month is genuinely useful for evaluation rather than a token allowance — as the arithmetic below shows, it covers hundreds of CVs. And concurrency, not price, is what separates Starter from Pro: five concurrent parse jobs on both Free and Starter, twenty on Pro, one hundred on Enterprise. If you are parsing a batch of CVs on upload rather than one at a time, that limit will reach you before the credit ceiling does.

Credits per page, by tier

Add-ons are priced the same way: layout extraction is +3 credits a page on any tier, enriched forms output is +10 credits on each page containing a form, spreadsheets bill 1 credit a sheet, and audio 3 credits a minute. Caching is free — the docs list "Smart result caching (re-parse = 0 credits)" — which matters if your pipeline re-runs over the same document set.

A two-page CV on the Cost-effective tier therefore costs about two-thirds of a cent. That is a very low number, and it is the number most blog posts stop at. It buys you markdown.

The step the markdown price does not include

A CV in markdown is not a parsed resume. To get fields you either write the extraction yourself — prompts, a schema, validation, date normalisation, and the maintenance of all four — or you pay for LlamaIndex's Extract product, which is priced on top of the parse.

In the v2 API the extract tiers publish an all-in default that bundles both steps: Cost-effective at 8 credits a page, Agentic at 25, Agentic Plus at 60 (an extract rate of 5, 15 or 50 plus the parse tier underneath it). There is also a Turbo extract tier at 35 credits a page that prepares the document itself and so carries no separate parse charge. Schema size scales it further: the docs state a multiplier on the extract tier, ×2 from 251 fields, rising to ×5 at 3,001–3,200 fields. A resume schema sits comfortably in the ×1 band.

So the honest range for a field-level CV parse on LlamaParse is roughly two cents to fifteen cents a document, depending on the tier you trust for the accuracy you need — not the two-thirds of a cent the parse rate implies. On the Free plan's 10,000 monthly credits that is about 625 two-page CVs at Cost-effective extraction, or 200 at Agentic.

How that compares to a resume-specific API

Our own plans are published on the docs page: Basic is $0 for 100 parses a month as a hard cap, Pro has no monthly fee and bills $0.05 a parse, Ultra is $29/month for 1,000 parses (about $0.029 each) with $0.045 overage, and Mega is $99/month for 5,000 (about $0.0198 each) with $0.018 overage. A parse is a whole document, not a page, and what comes back is typed resume JSON against a fixed field reference — no schema for you to design, version or maintain.

Read that table as a fit question, not a price question. On the cheapest extraction tier LlamaParse is at or below our per-parse price, and if your schema work is already done, that is a real saving. What it does not include is the part of resume parsing that is neither parsing nor prompting: deciding which of three overlapping date ranges is the current role, normalising "Sr. SWE II" into a title, or keeping all of it stable when a CV arrives in a two-column template.

Who should stay with LlamaParse

Plenty of readers should, and the cost table is not the reason:

• You parse many document types, not just CVs. Invoices, contracts, filings, scientific papers — LlamaParse is built for exactly that spread, and running one parser across all of them beats stitching several vendors together.
• You are building a RAG or agent pipeline where the document's layout, tables and charts have to survive into the model's context. That is the product's stated purpose, and a resume parser throws all of it away by design.
• You want control of the schema. If the fields you need are unusual — a security clearance, a registration number, a taxonomy of your own — an extraction product you steer will get closer than any fixed resume schema.
• You need self-hosting or a specific data region. LlamaIndex documents both; a small hosted API generally does not.

Stay with them in those cases. Our claim is narrower: if the documents are CVs and the output you need is typed resume fields, a parse-only, per-document API is fewer moving parts and a flatter bill.

How to switch, if the fit says you should

1. Take twenty CVs that are representative of your real intake — including the ugly ones, the scans and the two-column templates.
2. Run them through whatever you have today and keep the output.
3. Parse the same twenty against a resume API. Ours needs no key to try: the free browser parser hits the same endpoint with smaller limits.
4. Compare on the fields you actually act on, not on the whole blob. Employer, title, start and end dates, and current-role detection decide most product behaviour.
5. Price the winner against your real monthly document count, then re-read both rate cards before you commit — see also our CV parser API comparison and, if you are weighing a document converter rather than a hosted parser, Docling alternative.

The comparison takes an afternoon and it settles the question with your documents rather than somebody's marketing page — including ours.