Official-source checks
Assessment families, score-report guidance, and passing standards are checked against current NBME and USMLE pages.
About NBMEcalc
An Independent Score Planning Tool
NBMEcalc combines practice inputs into a planning range. It is not an official score report, and it does not claim a validated cohort until the underlying dataset can be reproduced.
Honesty over hype
We show confidence intervals, not a single "magic" number. We tell you when our model is uncertain.
Independent of NBME
We build models from public conversion data and aggregated user-submitted scores. We are not affiliated with the NBME or USMLE.
Built for you, not advertisers
We will never sell your data. The only way we make money is by you choosing to upgrade — and we make sure that choice is worth it.
Algorithm transparency
The methodology and material changes are public. Claims without reproducible evidence are labelled as assumptions or removed.
Transparent About What the Model Can Do
The calculator exposes its assumptions and uncertainty. We do not claim a validated cohort until a reproducible dataset and holdout report are published.
0
Step Exams Supported
Step 1, Step 2 CK, and Step 3
0
Input Source Types
NBME, UWSA 1/2, Free 120, AMBOSS, and CMS
0%
Interval Reported
Planning range, not a guarantee
How we got here
A short, honest timeline. We'll keep updating this page.
- 2025 Q4
Idea & first prototype
Frustrated with closed-source predictors and one-size-fits-all advice, the founder built a multi-source predictor in a weekend.
- 2026 Q1
Public methodology
Model assumptions, official-source checks, limitations, and material changes are documented for every visitor.
- 2026 Q2
Public launch
Free tier opens to all. Single Report and Pro plans launch with PDF delivery, multi-Step tracking, and confidence intervals.
- 2026 H2
ML predictor v2
Once we collect 1,000+ real outcome pairs, we replace the linear model with a gradient-boosted tree and re-validate.
Editorial controls
Evidence and Review Status
NBMEcalc currently uses an internal editorial process. We do not claim named physician review until reviewer identities and scope can be verified publicly.
Versioned model changes
Material assumption changes are documented publicly. Stored predictions keep their algorithm version.
Corrections welcome
Readers can report factual or calculation concerns. Unsupported claims are corrected or removed rather than defended.
Read the public methodology and changelog or submit a correction.
Got feedback or a partnership idea?
We read every email. Reach the founder directly at hello@nbmecalc.com.