One-time pricing for training runs

Pick a plan that fits your dataset. We verify limits before training starts, so you only pay when your run is eligible.

No subscriptions. Clear limits. Download your artifacts.

Pricing at a glance

  • Plans start at $5 per training run.
  • Launch: smaller datasets, 4-day retention, download-only delivery.
  • Orbit: larger datasets, 7-day retention, optional API capability.

Limits are checked before training begins, so you won't be charged for failed validation.

Plans

Start with Launch for smaller runs. Choose Orbit when you need more room, longer retention, and optional API capability.

Launch S

$5per run

Includes training + downloadable artifacts.

  • Dataset rangeUp to 50 MB
  • Retention4 days
  • API accessNot included
Token cap: 5.5M
Max records/lines
200,000
Max line length
20,000 characters

Recommended for: Good fit for small JSONL datasets and fast first experiments.

Most common

Launch M

$7per run

Includes training + downloadable artifacts.

  • Dataset range50–150 MB
  • Retention4 days
  • API accessNot included
Token cap: 7.5M
Max records/lines
500,000
Max line length
20,000 characters

Recommended for: Balanced for early production-like iteration with moderate data volume.

Launch L

$9per run

Includes training + downloadable artifacts.

  • Dataset range150–300 MB
  • Retention4 days
  • API accessNot included
Token cap: 10.0M
Max records/lines
1,000,000
Max line length
20,000 characters

Recommended for: Best when you need full Launch capacity before moving to Orbit.

Already training with BeaverYard? Log in.

Quick plan finder

Know your dataset size? Find the plan that fits.

Not sure about tokens? Upload first and we'll estimate before you pay.

Not sure yet? Upload first and we'll estimate before you pay.

Enter dataset size (and optional tokens) to get an instant best-fit recommendation.

If you're not sure about tokens, upload first and we'll estimate before you pay.

On this page

Ready to start, or want one more quick recommendation?

Compare limits

All plans include one training run. Limits are verified before training begins. Most people won't need this. It's here for precise comparisons.

Optional add-ons

Add only what you need. Your total is always shown before you pay.

Priority Processing

Start sooner with reduced scheduling buffer.

Launch: +$2 / Orbit: +$3Choose before training

Scheduling only: faster queue priority, no change to model quality or limits. Buy before training.

Extended Artifact Retention

Keep your artifacts available for 30 days.

+$3Can buy after training

Extends download window to 30 days, not training quality. Buy at checkout or before expiry.

Training Logs Bundle

Download raw + structured logs and a config snapshot.

+$3Can buy after training

Adds raw + structured logs with config snapshot, not custom tuning support. Buy at checkout or after completion.

Educational Summary: Training Intent

A short explanation of what your run is doing and why.

+$2Can buy after training

Adds a plain-language intent summary, not model changes. Buy at checkout or after completion.

Educational Summary: Use & Limitations

Clear guidance on what the model is good at, and where it will struggle.

+$2Can buy after training

Adds strengths/limitations guidance, not production compliance validation. Buy at checkout or after completion.

Pay only when your run is eligible

  • We verify plan limits before training starts.
  • If validation fails, training won't start and you won't be charged.
  • You'll see the reason and what to change to try again.
Common reasons

SIZE_OVER_LIMIT

Choose a larger plan or reduce dataset size and try again.

TOKENS_OVER_LIMIT

You can upgrade, or approve trimming/downsampling to match the plan limit.

RECORDS_OVER_LIMIT

Reduce rows, split the dataset, or choose a larger plan.

LINE_TOO_LONG

Break long records into smaller parts and re-upload.

Common questions

Start Run

No subscriptions. Clear limits. Download your artifacts.

Start Run