For developers building retrieval-augmented apps

    Your corpus is a storage cost,
    not a per-query tax.

    S3-compatible storage at $4.99/TB flat. Store the whole document corpus without per-read or per-request fees eating into retrieval margin.

    Start for freeTalk to an expert

    No credit card required · No per-request fees · Connects in minutes

    Retrieval economics tax the corpus, not just the query.

    The storage layer for a retrieval-augmented pipeline looks cheap on the surface. Then the egress and per-request lines compound with query volume and the corpus size becomes something to prune.

    The corpus cap

    The budget decides how broad the corpus is.

    Every document stored has a per-GB cost. Every retrieval read has an egress cost. Teams start pruning the corpus — not because the documents have no value, but because the storage economics do not reward coverage.

    The retrieval bill

    Every retrieval hits the egress line.

    A 1 TB corpus read 5 times a month for batch retrieval incurs $450 in egress on AWS at $0.09/GB — on top of $23.55 in storage. The storage is the smaller charge. The reads are the bill.

    The scale wall

    More documents means higher cost per query.

    Add more documents, pay more per retrieval pass, pay more per API call. The corpus that would answer more questions costs more to maintain. Teams cap it at a size they can afford, not a size that is useful.

    Same S3 API. A corpus bill determined by storage, not reads.

    Any S3-compatible client reads and writes the document store. Retrieval reads are included in flat storage — no egress counter.

    corpus.py
    import boto3, os

s3 = boto3.client(
    "s3",
    endpoint_url="https://eu-west-1.s3.fil.one",
    aws_access_key_id=os.environ["FIL_ACCESS_KEY"],
    aws_secret_access_key=os.environ["FIL_SECRET_KEY"],
    region_name="eu-west-1",
)

# Store a document chunk for retrieval
s3.put_object(
    Bucket="rag-corpus",
    Key="documents/doc-001.txt",
    Body=document_text.encode(),
)

# Retrieve during augmentation — egress $0
obj = s3.get_object(
    Bucket="rag-corpus",
    Key="documents/doc-001.txt",
)
context = obj["Body"].read().decode()
    AWS S3 Standard$23.55 storage + $460.80 egress$484/mo
    Fil One1 TB × $4.99 — egress $0$4.99/mo

    AWS S3 Standard us-east-1 Q2 2026: $0.023/GB storage, $0.09/GB egress. Computed from stated inputs — 1,024 GB × $0.023 = $23.55 storage; 5,120 GB × $0.09 = $460.80 egress. Fil One: 1 TB × $4.99 = $4.99, egress $0.

    Storage that scales with corpus size, not query rate.

    The storage layer is live today. Native RAG tooling integrations are in development.

    Store the whole corpus

    At $4.99/TB flat, a 1 TB document store is $4.99/month. A 10 TB corpus is $49.90. The storage price does not penalise breadth.

    No egress on retrieval reads

    Retrieval pipelines read raw documents on every query. Those reads are included in flat storage — there is no $0.09/GB egress line on retrieval.

    Predictable corpus cost

    Storage volume times $4.99. No per-request fees on reads or writes. The corpus bill grows with what you keep, not how often you query it.

    Native RAG integrations

    In development

    LangChain, LlamaIndex, and Haystack connectors for direct corpus management. Join the waitlist for early access.

    One rate. $4.99/TB/month.

    Storage. That is the whole bill. No egress fees on retrieval reads, no per-request fees. The corpus cost is the TB you keep, multiplied by one number.

    Start for freeTalk to an expert

    No credit card required · No per-request fees · Connects in minutes

    Store the whole corpus.

    Free 1 TB evaluation. Point your existing S3 client at the endpoint and stop rationing coverage.

    Start for freeTalk to an expert

    No credit card required · No per-request fees · Connects in minutes