Custom models · your data · your weights Turn your data into
Turn your data into
your own AI model.
Fine-tune open-weight models on your data, serve OpenAI-compatible endpoints, and export the weights whenever you want. Metered, pay-as-you-go.
No card required · export your weights any time
studio.scalty.com
support-assistant-prod
Live requests / minmetered
POST /api/v1/chat/completions 200
The loop
CONNECT
PREPARE
TRAIN
INTERACT
SHIP
METER
Everything the loop needs
Connect any source
Upload JSONL/CSV, or pull from S3, Postgres, and more. Snapshot into versioned datasets.
Fine-tune open weights
LoRA on license-clean bases with guided defaults — or override every hyperparameter.
OpenAI-compatible
Serve stable named endpoints. The unmodified OpenAI SDK just works.
Ground with your docs
Attach knowledge bases for retrieval-grounded answers with citations.
Own your weights
Export adapters, merged safetensors, or GGUF. Run them anywhere — leaving is always allowed.
Metered, pay-as-you-go
Every figure comes from the usage ledger. $5 monthly credit, no plans, no seats.
API-first
Built for developers
Point the OpenAI SDK at your endpoint and go. Streaming, usage, and citations come back in the shapes you already handle.
from openai import OpenAI client = OpenAI( base_url="https://studio.scalty.com/api/v1", api_key="sk_live_…", ) r = client.chat.completions.create( model="support-assistant-prod", messages=[{"role": "user", "content": "Hi"}], )
Transparent, metered pricing
Pay only for what you run. $5 of usage credit every month, no card to start.
Training
LoRA fine-tuning
$0.05/min + $0.25 launch
Tokens
text inference
$0.20 in · $0.90 out / 1M
Predictions
tabular
$1.00 / 1,000 rows
Storage
datasets & artifacts
$0.10 / GiB-mo
Knowledge-base storage (vector index) is billed at $0.40 / GiB-mo. No subscriptions, no seats — every figure comes from your usage ledger.