open data

The open neobank dataset

381 verified-active neobanks · MIT-licensed · machine-readable · updated 2026-09-11

Everything on neobankbeat is generated from one file: data.json. It's free to use, including commercially — attribution appreciated. Defunct neobanks and pure BaaS/infrastructure providers are excluded; unverified fields are null rather than guessed.

381

neobanks

261

traditional

hybrid

web3-native

128

niche-audience

data.json — full dataset ↓ openapi.json — API spec ↓ llms.txt — agent guide ↓ GitHub — source & issues ↗

Field dictionary

Each entity is an object under entities. Top-level meta carries counts, field notes and methodology.

field	type	meaning
name	string	Display name of the neobank.
category	enum	traditional · hybrid · web3-native (see notes).
audience	enum	general, or a niche: SMB, teens, gen z, women, migrants, freelancers…
hq · founded	string · number	Headquarters (city, country) and year founded.
custody	enum	Custodial · Self-custodial · MPC self-custodial · Mixed.
region	enum	Primary market, exactly one value: US · Canada · UK · US/UK · Europe · LatAm · Asia · ANZ · Africa · MENA · Global. Distinct from active_regions, which lists everywhere it operates.
regulation_type	enum	Broker-led · E-money institution · Fintech license (CH) · License pending (partner model today) · Licensed bank · MiCA CASP (EU) · Other / mixed · Partner-bank model · Payment institution · Self-custodial software · VASP / MSB / crypto licenses.
license	string	Plain-text license / sponsor-bank detail.
card_network · card_type	string	Visa/Mastercard/domestic/— and debit/prepaid/credit/virtual/crypto-settled/wallet.
cashback · yield	string	Headline "up to" figures — change constantly, always confirm with the issuer.
stablecoins	boolean	Whether stablecoins are supported.
services	array	Money-movement capabilities: on-ramp, off-ramp, fiat-payin, fiat-payout, iban, multi-currency, virtual-cards, crypto-cards.
fx_markup	object	FX markup on the free/standard plan, with source URL + as_of date.
kyc	string	KYC posture.
active_regions · countries	array	Macro-regions and country-level availability where known.
reported_users	object	value_millions + metric + as_of, where disclosed.
founders · funding · investors	string · array	Named founders, total disclosed funding, and notable disclosed backers.
ai	enum	Verified in-production AI use: underwriting · interface · agentic.
volume	object	figure + metric + source, for the few that disclose payment or transaction volume rather than user counts.
note	string	One line of current, factual detail — scale, license status, a recent launch.
story	string	One line on how the company got here. Editorial, not a data field: use it for context, not for analysis.
website · domain · x_handle · terms_url · privacy_url	string	Verified links where confirmed; null when unverified (never fabricated).
slug	string	Stable URL slug — the /n/<slug>/ profile path. Derived from name, persisted so it never drifts.
features	object	Tri-state capability flags (true / false / null=unverified): apple_pay, google_pay, api, webhooks, oauth, sandbox, business_accounts, virtual_cards, physical_cards, iban, self_custody, mpc, yield_bearing, savings, lending, investing, treasury, crypto, stablecoin_native. null is never "no".
stablecoin_tickers	array	Which stablecoins are supported (USDC, USDT, EURC…); [] = verified none, null = unverified. Companion to the stablecoins boolean.
card_networks	array	Normalized card networks (Visa, Mastercard, Amex, RuPay, Verve, Mada…); supersedes the free-text card_network.
card_form · card_funding	array	Orthogonal card axes: form (virtual, physical) and funding (debit, credit, prepaid, charge, bnpl); split out of card_type.
payment_rails	array	Instant/interbank rails (SEPA, SEPA-Instant, FedNow, ACH, Wire, SWIFT, Pix, UPI…); null = unverified.
fx_markup_pct · cashback_pct · yield_pct	number	Numeric companions to the fx_markup / cashback / yield prose — headline percent; null when no number is stated.
timeline	array	Dated events {date, type, description, source, tags}: founded, funding, launch, license, expansion, milestone… null = none recorded.
partners	object	Infrastructure map: issuer, partner_bank, treasury, identity, kyc_provider, card_network — who actually holds and moves the money. null = unverified.
funding_stage	enum	Investor-lens stage: Pre-seed / Seed / Series A / B / C+ / Public / Acquired / Bootstrapped / Subsidiary. Heuristic — explicit signals (public ticker, "acquired", named round) are reliable; the $-amount fallback is directional. null = undisclosed.

Notes & methodology

category: traditional = licensed fiat, custodial · hybrid = fiat + custodial crypto · web3-native = self-custodial, on-chain-first
rates: cashback/yield are 'up to' figures that change constantly and vary by region — always confirm with the issuer
services: money-movement capabilities (on-ramp, off-ramp, fiat-payin, fiat-payout, iban, multi-currency, virtual-cards, crypto-cards); verified per provider docs, tags omitted when unverified — absence is not proof of absence
fx_markup: foreign-exchange markup on the free/standard plan, sourced + dated; rates change often, always confirm with the issuer
verification: defunct neobanks and pure BaaS/infrastructure are excluded; unverified fields are null rather than guessed
graveyard: delisted entities are archived in the graveyard array with status 'delisted', the delisting date and a cause of death (kind: rail = partner/rail exit, acquired, regulator) — never silently deleted; excluded from all counts

Load it

import json, urllib.request
data = json.load(urllib.request.urlopen("https://www.neobankbeat.com/data.json"))
print(data["meta"]["total"], "neobanks")
for e in data["entities"][:5]:
    print(e["name"], "·", e["category"], "·", e["regulation_type"])

Mirrors

The same data is published on the two hubs researchers and model pipelines actually search. Both are refreshed from this file, so data.json stays the source of truth — if the three ever disagree, this one is right.

Hugging Face — neobankbeat/neobanks ↗ Kaggle — neobankbeat/neobanks ↗

The mirrors add two shapes this site doesn't serve: entities.jsonl, one neobank per line, and entities.csv, flattened to plain columns with the nested fx_markup, reported_users and volume objects split into their own _as_of and _source fields. Both download without an account:

from datasets import load_dataset

ds = load_dataset("neobankbeat/neobanks", split="train")
print(ds.filter(lambda e: e["regulation_type"] == "Licensed bank").num_rows, "hold a bank license")

Or straight to the flat table: entities.csv · entities.jsonl

Cite it

neobankbeat (2026). Open directory of neobanks worldwide. https://www.neobankbeat.com/ (MIT).

go deeperBrowse the interactive directory, all 381 profiles, or the comparisons. Machine agents: llms.txt · sitemap.md.

using this at work?Teams at fintechs, VCs and banks use this data for market maps, diligence and monitoring. For sponsorships, data licensing or a monthly team briefing — work with us →

Figures compiled from public sources, for comparison only — not financial advice. "Up to" rates change constantly; always confirm with the issuer. Spotted an error? Suggest a fix.