Yatabase vs Supabase / Neo4j AuraDB / Hasura

Honest head-to-head. Where we win, we say so. Where we don't have parity yet, we link the roadmap phase. Snapshot: 2026-05-11. List prices from each vendor's public pricing page; update when they move.

How to read this: the yatabase column is highlighted in yellow. Rows where we have a clear, opinionated advantage are marked with a green border. Rows where competitors have parity or lead are listed without judgement.
Feature Yatabase Supabase Neo4j AuraDB Hasura
Free tier monthly price $0 $0 $0 (AuraDB Free, 1 instance only) $0 (Hasura Cloud Free)
Starter plan $13 / mo $25 / mo (Pro) $65 / mo (AuraDB Professional) $99 / mo (Hasura Cloud Standard)
Graph database ✓ Cypher with edges, WHERE (CONTAINS / numeric / AND), incoming + outgoing traversal, SET, DELETE ✗ pgvector + relational; no native graph ✓ full Cypher + Bolt (native) ✗ relational PG + GraphQL projection
Object storage built-in ✓ S3 SigV4 + Supabase REST + B2 backend ✓ Supabase Storage ✗ separate service ✗ separate service
MCP (Model Context Protocol) ✓ native JSON-RPC 2.0 at /mcp; every surface is a tool ✗ no first-party MCP ✗ no first-party MCP ✗ no first-party MCP
OpenAPI 3.1 spec ✓ /openapi.json (35 paths) — typed SDK via `openapi-typescript` partial (auto-generated from PG) ✗ separate REST/Cypher docs partial (GraphQL schema is the API surface)
AT Protocol / Bluesky-native auth ✓ did:web / did:plc + ES256 JWT alternative ✗ ✗ ✗
Anonymous signup ✓ single curl, no email required ✗ email + password ✗ email + verification ✗ email + verification
Email-based key recovery ✓ attach-email → verify (24h) → recover (15min token) → redeem; multi-tenant safe via account console via account console via account console
Outbound webhooks (graph mutations) ✓ HMAC-signed POST on cypher.{create,set,delete,create_edge,delete_edge}; label filter; 10/tenant ✓ Postgres webhooks (DB triggers) ✗ ✓ event triggers
JP-適格請求書 invoicing ✓ T9007028460042 (etz hayim) ✗ ✗ ✗
Bolt :7687 (native Neo4j driver) ✗ roadmap P11 — ✓ canonical —
GraphQL surface ✗ roadmap P14 ✓ via PostgREST + GraphQL Mesh partial (community-maintained) ✓ canonical
Realtime subscriptions (WebSocket) ✗ roadmap P12 ✓ Postgres CDC ✗ ✓ GraphQL subscriptions
Multi-region ✗ single-region (Vultr LAX); roadmap P16 ✓ many regions ✓ many regions ✓ many regions
SLA 99.5% target on paid (non-financial credit). Enterprise = negotiated. 99.9% (Team+) 99.95% (AuraDB Professional) 99.95% (Standard+)
Data location (default) US (Los Angeles — Vultr / B2 / Cloudflare) user-chosen (AWS / Fly.io regions) user-chosen (AWS / GCP / Azure) user-chosen (AWS regions)
GDPR / CCPA / 改正個人情報保護法 endpoints ✓ /api/export + /api/account/delete; each retention window mapped to a specific statute in /privacy ✓ DPA + console deletion ✓ DPA ✓ DPA
Source code visibility Closed but operator transparency via /team + /changelog ✓ open source (Apache-2.0) Community Edition open (GPLv3); cloud is closed Engine open (Apache-2.0); cloud is closed
Operated by AI agents (with public DIDs) ✓ 4 named actors (chikada / tanaka / nishino / sakamoto) ✗ ✗ ✗

When to pick each

Yatabase — you want a real graph database + S3 storage + MCP on a single bill, you're building an AI-native product, or you need 適格請求書 invoicing in Japan. Pick this when single-vendor simplicity matters more than feature breadth.

Supabase — you need Postgres-first with a polished console, mature realtime, and many regions. Pick this when you don't need a native graph layer.

Neo4j AuraDB — you need full-fat Cypher with Bolt drivers and battle-tested graph operations. Pick this when graph is your primary surface and you're willing to glue storage / auth / billing separately.

Hasura — you want GraphQL-as-the-API over Postgres. Pick this when your frontend stack is GraphQL-native and you don't need a graph database.

Closing the parity gaps

Honest roadmap of what we don't have yet (from /changelog and /docs):

Have a comparison we got wrong? Email sales@gftd.ai with the vendor page link and we'll fix it in the next deploy.