Cheaper than the interview hour it replaces.
A full assessment — task, workspace, CI on every commit, and an evidence-linked scorecard — costs a fraction of one senior-engineer interview hour. The token budget is the marginal cost. Pay per assessment; no seats, no platform minimum.
Screen
Async coding screens for pipeline volume.
- 2–4 hours of work, 48-hour window
- Model gateway + git remote
- CI on every commit
- Auto-scored report per candidate
- Token budget included
Loop
Full-depth loops for final rounds.
- Everything in Screen
- Live proctored sessions
- LLM-judged evaluation with rubric
- Session replay for reviewers
- Custom models and budgets
- ATS integration
Org
One bar across the whole company.
- Everything in Loop
- Task library from your incidents
- Calibration analytics by level
- SSO and audit exports
- Dedicated support
An assessment is one candidate running one task: session credentials, the metered model budget, CI on every push, and the scored session record. Unused invites never expire.
What's in each tier.
| Screen | Loop | Org | |
|---|---|---|---|
| Model gateway + private git remote | ✓ | ✓ | ✓ |
| CI on every commit | ✓ | ✓ | ✓ |
| Token budget included | up to 1.5M | custom | custom |
| Session replay + transcript | summary | ✓ full | ✓ full |
| LLM-judged rubric scoring | standard rubric | ✓ your rubric | ✓ your rubric |
| Live proctored sessions | ✗ | ✓ | ✓ |
| ATS integration | ✗ | ✓ | ✓ |
| Task library from your incidents | ✗ | ✗ | ✓ |
| Calibration analytics by level | ✗ | ✗ | ✓ |
| SSO + audit exports | ✗ | ✗ | ✓ |
Your first cohort is on us
Run up to five assessments against your own repo, free — including the model budget. If the session records don't change how your team debriefs, walk away.
Start a pilot →50% off for early-stage teams
Pre-Series A companies hiring their first ten engineers get half off every tier. The first hires set the bar — measure them properly.
Check eligibility →Where the money goes.
The token budget is the cost
A full 1.5M-token assessment runs roughly $7–30 in model spend depending on the model mix — and that's the marginal cost per candidate. The platform fee covers the remote, the CI, the replay, and the scoring.
The alternative
Four senior engineers × two hours each for a whiteboard loop costs more than a hundred Taste assessments — and leaves you with notes instead of evidence.
Pricing questions, answered.
What counts as an assessment?
One candidate running one task. It includes the session credentials, the metered model budget, CI on every push, and the scored session record. An invite you never send — or a candidate who never starts — doesn't count.
Is the model spend included in the price?
Screen includes a budget of up to 1.5M tokens per assessment. Loop and Org set custom budgets; model spend passes through at cost from your configured providers — you can even bring your own provider keys.
What if a candidate doesn't finish?
The session record still exists — partial work is often the most informative signal. Sessions abandoned before the first model call or push are not billed.
Do you do annual contracts?
Yes — volume and annual pricing on Loop and Org. Talk to us about your hiring plan and we'll price against it.
Decide on proof, not vibes.
Book a demo and run a sample task against your own repo in under a day.