LlamaIndex Free Plan: What You Actually Get Before Paying
The LlamaIndex free plan pairs a free MIT-licensed framework with 10,000 free LlamaCloud credits a month. Here is what each tier includes, costs, and limits.
By Galactic News Editorial · Published 2026-11-11T10:20:37+00:00 · Updated 2026-11-11T10:20:37+00:00
LlamaIndex Free Plan: What You Actually Get Before Paying
The LlamaIndex free plan gives you two free things. The open-source framework costs nothing under the MIT license. The managed LlamaCloud platform gives you 10,000 free credits every month. That is the short answer. The details matter, because the two free tiers work very differently. This guide explains what each one includes, what it leaves out, and when you will need to pay.
Two Free Products, Not One
LlamaIndex is both a company and a product name. The name covers two separate things. Confusing them is the most common mistake new users make.
First, there is the LlamaIndex framework. It is an open-source Python and TypeScript library for building retrieval-augmented generation apps and agents over your own data. It is free under the MIT license. You install it with pip or npm. You run it on your own machine. You pay only for the language model and vector database services you plug into it.
Second, there is LlamaCloud. It is the company's managed platform for document parsing and extraction. Its flagship tool is LlamaParse. It turns messy PDFs, Office files, and images into clean, structured data. LlamaCloud is a commercial service. It runs on credits. It has a free tier.
You can use the framework without ever touching LlamaCloud. You can also use LlamaParse without the framework. Most beginners should start with the free framework. Add LlamaCloud only when you need managed parsing at scale.
What the LlamaIndex Free Plan Includes
The LlamaCloud free tier is a pay-as-you-go plan with no subscription. The official pricing page lists 10,000 free credits per month. It charges $1.50 per 1,000 credits after that. It allows 100 users on the plan. Support is community level.
Basic parsing costs as little as 1 credit per page. At that rate, 10,000 credits cover roughly 1,000 pages of basic parsing per month. The official FAQ states this figure directly. Complex documents cost more. Layout-aware agentic parsing uses more credits per page than basic parsing.
| Feature | Free (Pay as you go) | Pro | Enterprise |
|---|---|---|---|
| Monthly price | $0 | See pricing page | Custom |
| Included credits | 10,000 | 400,000 | Custom |
| Credit price | $1.50 per 1,000 | $1.25 per 1,000 | Volume discount |
| Concurrent parse jobs | 5 | 20 | 100 |
| Projects | 1 | 5 | Custom |
| Users | 100 | 100 | Custom |
| Support | Community | Priority Slack | Dedicated |
The free tier includes the full parsing feature set. That covers advanced table extraction, chart and graph extraction, layout detection, structured JSON output, and more than 80 supported languages. It supports over 130 file types, including PDF, Office documents, spreadsheets, and images. Output formats include Markdown, plain text, JSON, XLSX, and HTML tables.
How LlamaCloud Credits Work
Credits are the currency of LlamaCloud. Every parse, extract, classify, and split job spends credits. The cost per page depends on the parsing mode you choose.
Basic parsing is the cheapest mode. It costs as little as 1 credit per page. Agentic parsing uses language models for layout-aware extraction. It costs more per page but handles complex documents better. The official pricing page offers an Auto Mode that routes each page to the cheapest suitable tier. The vendor says this can save up to 80 percent of credits.
Smart result caching also saves credits. Re-parsing a document you already parsed costs zero credits. Cached data is kept for 48 hours, then deleted. You can turn caching off if you prefer.
Do the math before you build a pipeline. A thousand simple pages a month fits inside the free tier. Ten thousand pages does not. Test your real documents on the free tier first. Then you will know your true cost per page.
What the Free Plan Does Not Include
The free tier is generous, but it has limits. Know them before you commit to it.
- One project only. The free tier allows a single project. Pro allows five. Teams with several apps will outgrow it fast.
- Five concurrent parse jobs. Heavy batches queue up. Pro allows twenty concurrent jobs.
- Community support. You rely on docs and community channels. Priority Slack support starts on Pro.
- No enterprise features. Single sign-on, virtual private cloud deployment, custom agreements, and a dedicated account manager are Enterprise only.
- No credit card, no commitment. This is a benefit, not a limit. The free tier needs no card to start.
The framework itself has no such limits. It is open source. You can run it anywhere, at any scale, for free. Your costs there come from the model and database providers you choose.
How to Start on the Free Plan in 5 Steps
- Install the framework. Run
pip install llama-indexfor Python ornpm install llamaindexfor TypeScript. - Build a small retrieval app. Index a few of your own documents. Query them. Learn the core concepts before you spend anything.
- Create a LlamaCloud account. Sign up on the LlamaCloud console. The free tier starts with 10,000 credits a month.
- Parse a real document. Upload a messy PDF from your own work. Compare LlamaParse output with a basic parser. Check the credit cost per page.
- Decide if you need more. If 10,000 credits cover your volume, stay free. If not, check the paid tiers on the official pricing page. Paid plans start at $50 a month, according to the official FAQ.
Bottom Line
The LlamaIndex free plan is genuinely useful. The framework is free forever under the MIT license. The LlamaCloud free tier gives you 10,000 credits a month, enough for about 1,000 pages of basic parsing. That is plenty for learning, prototypes, and small projects. You will outgrow it when you need multiple projects, faster concurrent parsing, or priority support. Until then, start free, measure your real credit use, and pay only when the numbers say so.