LlamaIndex

Included quota

OSS framework: free (Python/TS, 160+ LlamaHub connectors, indexing/query/agent engines). LlamaCloud Free: 10,000 credits/mo (~1,000 pages), agentic OCR, structured extraction, 1 user, 1 project, 5 indexes, 50 files/index, basic support. Starter ($50/mo): 40,000 credits, 5 users, 50 indexes, 250 files/index, PAYG to 400k. Pro ($500/mo): 400,000 credits, 10 users, 5 projects, 100 indexes, 1,250 files/index, Slack support, 5x rate limits, PAYG to 4M. Enterprise: custom credits, unlimited users/projects/indexes, SSO, VPC, dedicated support.

What is public

LlamaIndex (llamaindex.ai - dual model: a free open-source RAG/agent framework in Python/TS + the managed LlamaCloud platform for document parsing/extraction/indexing) prices LlamaCloud on CREDITS (1,000 credits = $1.25, current; $1.00 was the older rate). Four tiers, all with access to every module (Parse, Extract, Classify, Split, Index): Free (10,000 credits/mo, ~1,000 pages, 1 user/project), Starter ($50/mo, 40,000 credits, 5 users, PAYG to 400k capped ~$500), Pro ($500/mo, 400,000 credits, 10 users, Slack support, 5x rate limits, PAYG to 4M capped ~$5,000), and custom Enterprise (SSO, VPC, dedicated support; ~$30K/yr AWS Marketplace example). The OSS framework is free; LLM/embeddings/vector-DB costs are separate.

Billing mechanics

The framework is free OSS (you pay your own LLM/embedding/vector-DB/infra). LlamaCloud runs on credits ($1.25 per 1,000; ~$0.00125 per credit): every operation - parsing, extraction, indexing, classification - consumes credits, and the per-page cost scales with complexity (Fast/Basic ~1 credit/page -> Cost-effective ~3 -> Premium/Agentic ~45-60 credits/page; LlamaExtract Premium = 60 credits/page). The Starter ($50/mo) and Pro ($500/mo) subscriptions equal their included-credit value at $1.25/1,000 (40k and 400k credits); beyond the allotment you pay-as-you-go at the same rate (capped ~$500 Starter / ~$5,000 Pro). A 48-hour parse cache makes re-parsing the same file free.

Variable cost rationale

Fully consumption-based - LlamaCloud spend scales with pages processed x complexity tier (credits), and the separate LLM/embedding/vector-DB costs scale with usage too; the parse cache and tier choice are the main levers

Additional watchouts

Credit-based pricing is hard to budget (complexity-driven swings + separate LLM/embedding/vector costs); the OSS path requires significant engineering time to manage infra/scaling; not built for non-technical business users; LlamaCloud only handles parsing/indexing - you still assemble the rest of the stack

Overage / add-ons

Beyond a tier's included credits, you pay-as-you-go at the same $1.25/1,000 rate - capped at ~$500 (Starter) and ~$5,000 (Pro); Enterprise negotiates custom credits + volume discounts. Per-operation credit cost varies sharply by parsing/extraction complexity tier; the 48-hour parse cache zeroes out repeat parses of the same file.

Sales call required

Mixed (some tiers require a call)

Free / trial

Free (10k credits/mo)

Lowest paid plan

n/p (credit-based platform)

Commercial notes

Formerly GPT Index; the leading RAG-specialized framework (vs LangChain's broader orchestration) - LlamaParse/LlamaExtract are its document-parsing moat; 160+ LlamaHub connectors; used by 10,000+ production teams; LlamaCloud is the managed layer (parse/extract/index) with a credit model; competes with LangChain (general orchestration) and no-code RAG tools (Stack AI, which bundles LLM tokens)

Key ambiguities

Resolved: the current credit rate is $1.25/1,000 (LlamaIndex's own pricing page + dev FAQ; $1.00/1,000 was the older rate). Starter ($50/mo) and Pro ($500/mo) are fixed monthly subscriptions whose price equals their included-credit value at $1.25/1,000 (40k / 400k credits), with PAYG beyond the allotment (capped per tier). Exact per-operation credit costs still vary by parse/extract complexity tier.

Cancellation / refund

OSS is free (no contract); LlamaCloud Free is free; Starter/Pro are self-serve with included credits + capped PAYG; Enterprise is custom-contracted (SaaS or VPC; ~$30K/yr AWS Marketplace example); credits reset monthly

Support SLA / resale

Basic support (Free/Starter); Slack support (Pro); dedicated support (Enterprise); SaaS or VPC deployment for enterprise; Enterprise SSO; Python/TS SDKs + REST API + vector-store integrations; available via AWS Marketplace

Missing data

Resolved: credit rate is $1.25/1,000 (current; $1.00 older); Starter $50/mo and Pro $500/mo are fixed subscriptions (= included-credit value) + capped PAYG. Exact per-operation credit costs vary by parse/extract tier. Seed 'Free tier 10k credits/mo (LlamaCloud); OSS framework self-hostable' is accurate and current; LlamaCloud is credit-based (Free/Starter/Pro/Enterprise) and LLM/vector costs are separate.