Free rag app stack
Answer questions over your own documents: an LLM, embeddings in a vector store, an API host and a bucket for the files.
Updated 20h ago
4 roles4 with no end date2 of 4 need no card
01 LLM API
Generates the answer from retrieved passages.
Google
Gemini API free tierFlash models, free of charge
No cardNo end date
- PriceFree of charge input and output (Flash)
- Rate limitsRPM, TPM and RPD per project
- ResetDaily quota resets at midnight Pacific
Swap LLM API
02 Vector store
Holds embeddings and finds the nearest passages.
Qdrant
Qdrant Cloud Free tier1 GB RAM cluster, forever
Card ?No end date
- ClusterSingle node
- Size0.5 vCPU, 1 GB RAM, 4 GB disk
- InferenceFree on selected models
Swap Vector store
03 API hosting
Runs the retrieval and prompt code.
Cloudflare
Workers Free plan100,000 requests a day
No cardNo end date
- Requests100,000 / day
- CPU time10 ms per invocation
- KV reads100,000 / day
Swap API hosting
04 Document storage
Keeps the source files you index.
Cloudflare
R2 free tier10 GB, zero egress fees
Card ?No end date
- Storage10 GB-month / month
- Class A ops1 million / month
- Class B ops10 million / month
Swap Document storage
Limits side by side
LLM API
Google Gemini API free tier
PriceFree of charge input and output (Flash)Rate limitsRPM, TPM and RPD per projectResetDaily quota resets at midnight PacificData useFree-tier content used to improve Google productsBatch, FlexNot available on free tierSign up
Vector store
Qdrant Cloud Free tier
API hosting
Cloudflare Workers Free plan
Requests100,000 / dayCPU time10 ms per invocationKV reads100,000 / dayKV writes1,000 / dayKV storage1 GBSign up
Document storage
Cloudflare R2 free tier
Storage10 GB-month / monthClass A ops1 million / monthClass B ops10 million / monthEgressFreeSign up
Offers change often; the vendor's page is the final word. Links go straight to the vendor, no affiliate links. Logos via logo.dev; trademarks belong to their owners.