Simple, transparent pricing

Generate high-quality synthetic data for LLM training.
Pay only for what you use, or save with a subscription.

Subscription Plans

Best value for regular usage. All plans use premium-tier models.

Starter

$49/mo

3,000 records/month

35% off vs pay-as-you-go

  • 3,000 premium records/month
  • All schemas (instruction, preference, eval)
  • Powered by Claude Sonnet 4.6
  • API access
  • Email support
Start Free Trial
Most Popular

Pro

$149/mo

10,000 records/month

40% off vs pay-as-you-go

  • 10,000 premium records/month
  • All schemas (instruction, preference, eval)
  • Powered by Claude Sonnet 4.6
  • API access
  • Priority support
  • Bulk generation
Start Free Trial

Scale

$399/mo

30,000 records/month

47% off vs pay-as-you-go

  • 30,000 premium records/month
  • All schemas (instruction, preference, eval)
  • Powered by Claude Sonnet 4.6
  • API access
  • Dedicated support
  • Priority job queue
  • Higher rate limits
  • 90-day data retention
Contact Sales

Pay-as-you-go

No commitment. Generate records on-demand at any quality tier.

TierModelinstruction_v1preference_v1eval_v1
Economy
Fast & affordable for experiments
GPT-4o Mini$0.50/1K—$0.50/1K
Standard
Balanced quality and cost
GPT-4o$5.00/1K—$5.00/1K
Premium
Highest quality for production
Claude Sonnet 4.6$25.00/1K$30.00/1K$25.00/1K

preference_v1 requires Premium tier for quality guarantee (100% acceptance rate)

Schema Availability

Each schema is optimized for specific ML training paradigms.

instruction_v1

Instruction-response pairs for supervised fine-tuning (SFT)

EconomyStandardPremium

preference_v1

Paired responses with preference labels for RLHF/DPO

Requires Premium tier for quality guarantee

EconomyStandardPremium

eval_v1

Evaluation datasets for benchmarking models

EconomyStandardPremium

Ready to get started?

Create your free account and start generating synthetic data in minutes.

Start Free TrialView Documentation

Frequently Asked Questions

Subscriptions give you a monthly allocation of premium-tier records at a significant discount (35-54% off). Pay-as-you-go lets you generate records on-demand at any tier without commitment. Subscriptions are best for regular usage, while PAYG is ideal for occasional needs or trying different tiers.
Preference data for RLHF/DPO training requires high-quality chosen/rejected response pairs. Our testing showed that cheaper models produce pairs that fail quality validation 50-60% of the time. Premium tier (Claude Sonnet 4.6) achieves 100% acceptance rate, ensuring you get usable training data.
One record is one complete data item in your chosen schema. For instruction_v1, that's one instruction-response pair. For preference_v1, it's one prompt with chosen and rejected responses. For eval_v1, it's one evaluation example.
Yes! Your subscription gives you premium-tier records. You can additionally purchase Economy or Standard tier records via pay-as-you-go for less quality-sensitive use cases.
You can continue generating at pay-as-you-go rates, or upgrade your plan for more monthly records. We'll notify you when you're approaching your limit.