Storage that keeps your GPUs fed
Store training datasets, model weights, and checkpoints with verifiable integrity — and move data freely with zero egress fees.
1 TB free for 30 days · No credit card required · No egress fees
Everything your training pipeline needs
Petabyte-scale dataset storage
Store training corpora, tokenized datasets, and raw crawl data with no practical size limits. Organize with prefixes and lifecycle rules.
High-throughput reads
Sustained multi-Gbps throughput so your GPU cluster never starves waiting for the next batch. Designed for continuous streaming reads.
Verifiable data integrity
Every object is cryptographically sealed. Prove your training data hasn't drifted or been tampered with.
No egress fees, ever
Move data between cloud regions, download checkpoints for fine-tuning, or replicate across providers. Zero egress charges, always.
S3-compatible, drop-in
Works with PyTorch DataLoader, HuggingFace datasets, Ray Data, and any S3-compatible SDK. One endpoint swap and you're live.
Checkpoint versioning
Tag model checkpoints with metadata, roll back to any point in training, and share artifacts across teams via presigned URLs.
Four steps to zero egress bills
Create a bucket
Provision a bucket in seconds from the dashboard. Pick your preferred region.
Swap your endpoint
Replace your existing S3 endpoint URL with Fil One's. No SDK changes, no re-architecture.
Upload your data
Use rclone, the AWS CLI, or any S3 library to migrate datasets and checkpoints.
Train with confidence
Your data is verifiably intact, your reads are fast, and your egress bill is zero.
Frequently asked questions
Each dataset is assigned a unique digital fingerprint (CID) at upload, creating a verifiable record of its contents. The system then automatically checks your data on a recurring basis (about every 24 hours) to confirm that the data still matches that original fingerprint, ensuring nothing has been altered.
Fil One provides an S3-compatible API: if your application works with AWS S3, it works with Fil One. Point your SDK or CLI at our endpoint and authenticate with your API keys.
Fil One takes a security-first approach to S3-compatibility. The S3-compatible API enables simple setup and migration. Storage supports private buckets by default, with public access consistent with full S3 parity coming soon.
Read Fil One docs, access the app to get started with no code required, or talk to someone on our team to get started.
Fil One uses industry-standard encryption with per-object data encryption keys, ensuring your data is protected regardless of which provider it's stored with. We are actively pursuing SOC 2 Type II and ISO 27001 certifications — if you'd like to be notified when those are finalized or have other questions about security, reach out to our team directly at security@fil.one.
After you create an account and provision your first bucket, migrating to Fil One is straightforward with our S3-compatible interface. You can integrate directly into your codebase using any S3-compatible SDK, or use the AWS CLI to migrate your data with a single command.
For enterprise teams who want more hands-on support during migration, our team is available to walk you through every step. Reach out at sales@fil.one, and we'll make sure the transition is seamless.