Four tiers, published in full, so you can budget before the first conversation. Every tier carries the same honesty system.
The honesty system is not a tier feature. Every plan gets the badges, the caps and the retired-claims log, because charging for the truth would defeat the point of publishing it.
| Capability | Explorer | Professional | Institutional | Enterprise |
|---|---|---|---|---|
| Asset verticals (built · real data) | All 14 | All 14 | All 14 | All 14 |
| API credits per month | 500 | 2,500 | Unlimited | Unlimited |
| Cited analyst | ✓ | ✓ | ✓ | ✓ |
| Morning brief & cited memos | Brief only | ✓ | ✓ | ✓ |
| =LATAVIS.* formulas + Excel add-in (pilot) | — | ✓ | ✓ | ✓ |
| Model risk documentation (SR 26-2 §VII) | — | — | — | ✓ |
| Honesty badges on every number | ✓ | ✓ | ✓ | ✓ |
What the API enforces today: each plan's monthly API credits. The other rows describe what your agreement includes; the API does not yet switch those features on or off by plan.
How API keys, rate limits and credits work, endpoint by endpoint: the API reference.
No, and we will not describe it as one. A terminal spans every asset class and a trading workflow. Our lead product is private credit, read from primary filings, with the coverage roster published so you can see the boundary before you sign rather than after.
No free tier and no trial. We offer a paid, scoped evaluation: an agreed set of filers or questions, a fixed fee and a fixed end date. The published validation record and the retired-claims log are public and ungated, so you can read them before you ask.
The private-credit disclosed-impairment score: leave-one-BDC-out AUC 0.5892 [95% CI 0.53, 0.65] on disclosed non-accrual (as of 2026-11-09). It ranks disclosed impairment; it is not a probability of default. No US recession probability is shown today: no recession model has passed its validation test. The model locked on 2026-10-15 was retired on 2026-11-01 (worse than the long-run recession rate on its settled outcomes, and two of its inputs are licence-held), and its licence-clean replacement failed its pre-registered test the same day (not significantly better than the long-run rate). The accuracy page also lists research records for models that are not offered in the product today; they are not part of any tier.
No. Verification badges, confidence caps, provenance and the retired-claims log are identical on every tier, including Explorer. Tiers change how much of the platform you can use, never how truthful it is.
The order matters. Read the validation record and the four withdrawn claims, and only then decide whether the price above is worth a conversation.