Skip to main content
JSONL with prompt / response pairs is preferred. CSV and plain text are accepted; LLMTune will prompt you to map columns.
How long does training take?
LoRA / QLoRA runs typically complete in 20–40 minutes depending on dataset size and GPU allocation. Full fine-tunes can take longer; the studio provides estimates before you launch.
Can I bring my own model checkpoints?
Yes. Enterprise plans allow import of custom base models. Contact support to connect storage and complete compliance review.
How are API requests billed?
You pay per input / output token processed by each deployment. Usage dashboards show daily and monthly totals. Training runs are billed per job based on GPU hours.
Can I export models?
You can download fine-tuned adapters or full-model weights (where licensing permits) from the deployment panel.
How do I monitor endpoints?
Use the Usage dashboards or subscribe to webhooks. You can also query /usage endpoints programmatically.
What safety features exist?
LLMTune supports guardrails such as redaction, toxicity classifiers, and custom filters. They can be enabled per training configuration and enforced at inference time.
How do I report an incident?
Use the in-app support panel or email [email protected]. Include workspace ID, deployment ID, and timestamps when possible.
What happens if I run out of balance?
Inference requests return 402 errors. Add balance via the Stripe integration in Usage → Balance and retries will succeed after credit is restored. 
