One-time pricing per run
No subscriptions, no monthly fees, no metered billing. One fixed price per training run, determined by your dataset.
Validate before you pay — failed runs cost nothing
Fine-tune any LLM.
One-time price.
Download your model.
Upload your dataset, get a free eligibility check, and pay only when your run qualifies. Adapter weights and plain-language guides ship with every completed run. No subscriptions. No hosted inference. Yours to deploy anywhere.
No subscriptionsPlans from $5LLaMA · Mistral · GemmaAPI access available
Simple workflow
From dataset to adapter in five steps
1
Upload dataset
JSONL format, any size within published limits.
2
Free validation
Format, size, token count, and record limits — checked before payment.
3
Review & pay
See your assigned plan and exact price. Pay only when eligible.
4
Training runs
We fine-tune using a standardized recipe with the base model you chose.
5
Download artifacts
Adapter weights and two customer guides. Yours within 7 days.
What makes BeaverYard different
Built around your workflow,
not ours
Validate before you pay
Format errors, size overages, token caps, record limits — all checked before we ask for payment. Failed validations cost nothing.
Download-first delivery
Your adapter weights and config files are yours to download and run on any compatible infrastructure. No platform lock-in.
Customer guides included
Every run ships with a Training Focus Summary and Usage Guidelines — plain-language docs on what your adapter learned and how to use it.
PEFT-compatible adapters
Artifacts ship as adapter_model.safetensors + adapter_config.json, ready for PEFT and popular inference frameworks out of the box.
API access available
API access is available on every plan. Integrate fine-tuning runs directly into your pipeline without touching the dashboard.
Deliverables
Everything you need after training
Three artifacts per completed run. Available to download for 7 days from completion.
Adapter package
- adapter_model.safetensors
- adapter_config.json
- PEFT-compatible format
- Run on any compatible host
Training Focus Summary
- What your adapter learned
- Behavioral changes to expect
- Recommended next steps
- Plain language, no jargon
Usage Guidelines
- How to load your adapter
- How to run inference tests
- Where it performs well
- Where it may struggle
Our commitments
Designed for transparency,
not complexity
Zero-risk validation
Validation runs before payment. If your dataset fails any limit check — format, size, token count, or record cap — you owe nothing. Not even a processing fee.
You own your artifacts
We don’t host your model, resell your training data, or lock you into a proprietary format. Your adapter weights run on any PEFT-compatible stack.
Fully transparent pricing
Plans are published. Assignment is automatic based on dataset size and token count. You see the final price before you pay — no negotiation, no hidden tiers.
Clear data handling
Data is processed in us-east-1 (AWS). Datasets and artifacts are automatically deleted after your 7-day retention window. Early deletion available on request.
Transparent pricing
Simple tiers. No surprises.
Plans are assigned automatically after upload and validation. The higher required tier wins when dataset size and token count disagree.
Launch
Datasets up to 300 MB · Up to 9.5M tokens
$5–$9/ run
- Adapter weights + config
- Training Focus Summary
- Usage Guidelines
- 7-day download window
- API access available
- One-time payment
Orbit
Datasets 300–600 MB · Up to 12M tokens
$15–$25/ run
- Adapter weights + config
- Training Focus Summary
- Usage Guidelines
- 7-day download window
- API access available
- One-time payment
Displayed ranges are planning bands. Final plan assignment happens automatically after upload and validation.
Common questions
Everything you need to know
BeaverYard is a fine-tuning service for open-source LLMs. You upload a JSONL dataset, we validate it against published limits before charging you, run the training job, and deliver downloadable adapter weights and customer guides.
Get started today
Ready to train your custom model?
Upload your dataset, get a free eligibility check, and pay only when your run qualifies. Plans start at $5.
No subscription required · No hidden fees · Validation is always free