Pricing

Custom eval suites, graded by real people, shipped in days.

Human-graded evals that catch what automated tests can’t

We build an eval suite around your product, then re-run and hand-grade it on every major update,
so you catch regressions before they ship. First suite in days, not quarters.

Suggested
Pilot
Prove it on one feature, risk-free
Suggested for
1LLM feature
  • One-time build and run
  • Up to 100 hand-graded case credits
  • Delivered in days
  • Failures ranked by severity
  • Credited toward month 1 if you continue
Book a call
Starter
Keep your LLMs ready for all cases
Suggested for · Up to
2LLM features
  • Monthly / Annual Retainer
  • Up to 200 hand-graded case credits
  • Graded on each major update (~2 runs/mo)
  • Flags anything that got worse since your last release
  • Monthly report
Book a call
Growth
Coverage across everything you ship
Suggested for · Up to
5LLM features
  • Monthly / Annual Retainer
  • Up to 1000 hand-graded case credits
  • Checks tailored to your failure modes
  • Per-release / weekly runs
  • Faster SLA
  • Shared Slack channel
Book a call
Scale
High-volume, audited, compliant
Suggested for
5+LLM features
  • Monthly / Annual Retainer
  • High run frequency
  • Dedicated grader pod
  • Tight SLA
  • Security / compliance review
Book a call
Fine print, the good kind
Get one month free on an annual retainer
Unused credits roll over and accumulate
No lock-in, cancel anytime
Who’s grading

Meet the human in human-graded

Pragyan Subedi

Pragyan
Subedi

Founder at Engineer In Residence
Know the founder

Pragyan is a senior machine learning engineer who’s shipped production models at Silicon Valley startups.

Vetted by Toptal
Top 3% Talent
Worked with datasets as large as
1B+ data points
Experience in data science
10+ years
Get started

Not sure which tier fits?

Most teams start with a Pilot on one feature, then move to a monthly retainer once they see what it catches. Book a call and we’ll scope the suite and the right tier for your product.

Book a call