AI responses, an independent measurement
Which database do AI assistants recommend
What 6 AIs recommend in this category
PostgreSQL leads overall, but the winner changes with the need.
Across the 12 situations tested, PostgreSQL is the tool most often recommended by the 6 assistants. But in 4 of 12 situations, the tool highlighted varies by AI. The choice therefore depends as much on the stated need as on the assistant used.
1
PostgreSQL
58%
2
MySQL
37%
3
SQLite
33%
4/12use cases where AIs disagree
430responses actually analyzed
The panelThe 14 compared tools View panel
ClickHouse
CockroachDB
DuckDB
DynamoDB
Firebase Firestore
MariaDB
Microsoft SQL Server
MongoDB
MySQL
Oracle Database
PostgreSQL
Redis
SQLite
Supabase
⓪ The big picture
Which tools do AIs recommend most often? ¶
The average recommendation rate is calculated across all tested situations, weighted equally. One assistant may recommend several tools, so these percentages are not expected to add up to 100%.
| # | Tool | Average recommendation rate | |
|---|---|---|---|
| 1 | PostgreSQL | 58% | |
| 2 | MySQL | 37% | |
| 3 | SQLite | 33% | |
| 4 | MongoDB | 33% | |
| 5 | Supabase | 33% | |
| 6 | CockroachDB | 27% | |
| 7 | Redis | 24% | |
| 8 | DynamoDB | 22% | |
| 9 | Firebase Firestore | 12% | |
| 10 | ClickHouse | 12% | |
| 11 | DuckDB | 3% | |
| 12 | MariaDB | 2% | |
| 13 | Microsoft SQL Server | 1% | |
| 14 | Oracle Database | 1% | |
① The answer depends on which AI you use
The tool each model spontaneously surfaces ¶
Use case by use case, the tool each model most often puts first (every use case counts the same, chatty or not).
claude-sonnet-5
→
CockroachDBleading on 2/12 use cases
deepseek-v4-pro
→
PostgreSQLleading on 8/12 use cases
gemini-3.5-flash
→
PostgreSQLleading on 8/12 use cases
mistral-medium-3-5
→
PostgreSQLleading on 7/12 use cases
gpt-5.2-chat
→
PostgreSQLleading on 5/12 use cases
grok-4.3
→
PostgreSQLleading on 5/12 use cases
Tool most often highlighted by each assistant: PostgreSQL (5) · CockroachDB (1)
② The models don't agree
For the same use case, a different dominant tool depending on the model ¶
Database for a SaaS startup MVP
consensus
Database for a side project
split
First database for a beginner learning backend development
consensus
Database for a mobile app backend
split
Database for analytics and reporting
consensus
Database for a high-traffic app at scale
consensus
Database for an enterprise app with strict compliance
consensus
Database for an e-commerce store
consensus
Database for a real-time collaborative app
split
Database for an AI app with vector search (RAG)
split
Embedded or local-first database for a desktop app
consensus
Database for a serverless or edge deployment
consensus
Database for a SaaS startup MVP
consensus
“Solid runner-up: PostgreSQL (self-hosted first, later managed) If you know you’ll need strict transactional integrity across many concurrent writers, or if your team already feels at home with Postgres, go with vanilla PostgreSQL.”
deepseek-v4-pro · gemini-3.5-flash · gpt-5.2-chat · grok-4.3
“Supabase (Postgres + Realtime)”
claude-sonnet-5 · mistral-medium-3-5
Database for a side project
split
“Supabase · PostgreSQL · Gives you a full‑powered Postgres with auto‑generated REST/GraphQL APIs, a nice web dashboard, and a free tier.”
deepseek-v4-pro · mistral-medium-3-5 · grok-4.3
“Supabase (The Best Overall for SaaS)”
gemini-3.5-flash · gpt-5.2-chat
“Plain SQLite is genuinely underrated for this scale.”
claude-sonnet-5
First database for a beginner learning backend development
consensus
“For a Django + React bootcamp, PostgreSQL is the best choice.”
claude-sonnet-5 · deepseek-v4-pro · gemini-3.5-flash · mistral-medium-3-5 · gpt-5.2-chat · grok-4.3
Database for a mobile app backend
split
“Where Document Stores (MongoDB, DynamoDB, Firestore) Win”
claude-sonnet-5 · deepseek-v4-pro
“Relational Store (e.g., PostgreSQL, Aurora)”
gemini-3.5-flash · mistral-medium-3-5
“Firebase Firestore (or Realtime Database) – Strongest all-rounder for indies”
grok-4.3
“✅ In-memory store (Redis)”
gpt-5.2-chat
Database for analytics and reporting
consensus
“For millions of events/day with sub-second aggregations, columnar storage (ClickHouse, Druid, BigQuery, Redshift, Snowflake) is almost certainly your answer.”
claude-sonnet-5 · deepseek-v4-pro · gemini-3.5-flash · mistral-medium-3-5 · gpt-5.2-chat · grok-4.3
Database for a high-traffic app at scale
consensus
“CockroachDB — Postgres wire-compatible, distributed writes, automatic sharding, full ACID with joins across shards. Migration is often "point your driver here" plus schema adjustments for hot-spotting.”
claude-sonnet-5 · deepseek-v4-pro · mistral-medium-3-5 · gpt-5.2-chat · grok-4.3
“Distributed SQL databases are designed to look and act like traditional relational databases (using PostgreSQL or MySQL wire protocols)”
gemini-3.5-flash
Database for an enterprise app with strict compliance
consensus
“You can achieve this with a battle-tested stack built around PostgreSQL (self-managed or cloud-hosted) combined with a few open-source tools and cloud-native immutable storage.”
deepseek-v4-pro · gemini-3.5-flash · mistral-medium-3-5 · grok-4.3
“4. CockroachDB If you need multi-region/HA down the line.”
claude-sonnet-5 · gpt-5.2-chat
Database for an e-commerce store
consensus
“Don't let "lightweight" push you away from Postgres — it's lighter on ops burden than people assume, especially with managed hosting.”
claude-sonnet-5 · deepseek-v4-pro · mistral-medium-3-5 · gpt-5.2-chat · grok-4.3
“If "real-time" means thousands of users clicking "Buy" at the exact same millisecond, Redis is the gold standard.”
gemini-3.5-flash
Database for a real-time collaborative app
split
“PostgreSQL with a partitioned event table (fast inserts, easy to query by document ID/timestamp). Use NOTIFY/LISTEN to push new ops to the live server, or poll with a head cursor.”
deepseek-v4-pro · gemini-3.5-flash · mistral-medium-3-5
“Consider a pub/sub layer (Redis Pub/Sub, NATS)”
claude-sonnet-5 · gpt-5.2-chat · grok-4.3
Database for an AI app with vector search (RAG)
split
“With pgvector you can implement RRF in application code, or use PostgreSQL’s full‑text search ranking (ts_rank) combined with vector distance, then apply a weighted sum or exponential decay.”
deepseek-v4-pro · gemini-3.5-flash · gpt-5.2-chat
“Managed Solutions (Pinecone, Weaviate Cloud, Qdrant Cloud, MongoDB Atlas Vector Search, etc.)”
claude-sonnet-5
“Extend the relational DB (e.g., pgvector, pgvecto.rs, MySQL Vector)”
grok-4.3
“Some managed services (Pinecone, Weaviate Cloud) offer built-in multi-tenancy; others (Redis, FAISS) require custom implementation.”
mistral-medium-3-5
Embedded or local-first database for a desktop app
consensus
“Start with plain SQLite (via better-sqlite3, Core Data, Room, etc.), structure your schema thoughtfully, and plan to add sync later via a separate mechanism.”
claude-sonnet-5 · deepseek-v4-pro · gemini-3.5-flash · mistral-medium-3-5 · gpt-5.2-chat · grok-4.3
Database for a serverless or edge deployment
consensus
“Neon (PostgreSQL over HTTP/WebSocket) - SQL flavour: Full PostgreSQL (with pgvector, extensions, etc.).”
deepseek-v4-pro · gemini-3.5-flash · mistral-medium-3-5 · gpt-5.2-chat
“Neon (Postgres) and PlanetScale (MySQL/Vitess) are the two most mature options:”
claude-sonnet-5 · grok-4.3
③ Each tool its territory
Which use cases AIs associate with which tool ¶
For each cell, the percentage shows the share of responses recommending the tool. Select a dot for details. ‘n’ is the number of responses analyzed; results based on few responses are indicative.
↔ Swipe the matrix to explore use cases.
| Tool | Database for a SaaS startup MVP | Database for a side project | First database for a beginner learning backend development | Database for a mobile app backend | Database for analytics and reporting | Database for a high-traffic app at scale | Database for an enterprise app with strict compliance | Database for an e-commerce store | Database for a real-time collaborative app | Database for an AI app with vector search (RAG) | Embedded or local-first database for a desktop app | Database for a serverless or edge deployment |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| ClickHouse | ||||||||||||
| CockroachDB | ||||||||||||
| DuckDB | ||||||||||||
| DynamoDB | ||||||||||||
| Firebase Firestore | ||||||||||||
| MariaDB | ||||||||||||
| Microsoft SQL Server | ||||||||||||
| MongoDB | ||||||||||||
| MySQL | ||||||||||||
| Oracle Database | ||||||||||||
| PostgreSQL | ||||||||||||
| Redis | ||||||||||||
| SQLite | ||||||||||||
| Supabase |
strongly recommended
cited
absent
Click a dot to show the detail of an association.
Your tool ranks poorly, or is missing? Submit it for the next wave →
Are you a brand? See what AI says about you →
④ The why
Why do AIs recommend each tool? ¶
The main argument the models invoke for each tool, the price they quote, and the caveat they attach.
ClickHouse
performance ×7
scalability ×3
aggregation speed
analytical workloads
Prices quoted: open-source · open-source (no license fees) · $ (self-hosted) · open-source, only pay for infrastructure; $100/month per VM example · open-source (no licensing fees) · free and open-source
Caveat: added complexity and operational cost, don't start there
CockroachDB
scalability ×38
strong consistency ×8
horizontal scalability ×7
scalability and consistency ×4
Prices quoted: serverless pricing · More expensive than Postgres at small scale · Free (self-hosted) or ~$0.10–$0.50/GB/month · Higher than Postgres · enterprise license (no figure given) · slightly more expensive than basic Postgres
Caveat: audit capabilities vary
DuckDB
scalability ×2
analytical queries
analytical speed
batch cost efficiency
Prices quoted: free / serverless · free · free (open-source)
Caveat: added complexity and operational cost, don't start there
DynamoDB
scalability ×25
consistency ×5
features ×5
horizontal scalability ×3
Prices quoted: inexpensive early · $0 (Pay-per-use) · pay-as-you-go · strong reads cost 2× RCUs · zero idle cost · pay-per-request
Caveat: eventual consistency
Firebase Firestore
simplicity ×15
better real-time experience (offline-first sync)
built-in offline sync and conflict resolution
compliance
Prices quoted: free tier · generous free tier · Free tier (Generous for prototypes: 50K reads/day, 20K writes/day); pay-as-you-go after · Free tier: 50k reads/day, 20k writes/day · Free tier: 50K reads/day, 20K writes · pay-per-document + generous free tier
Caveat: costs can climb unpredictably with chat apps, complex queries get awkward
MariaDB
features ×2
compatibility
existing setup
features (rls)
Prices quoted: Very cheap · ~$15/month
Caveat: audit logging requires mysql enterprise (pricey) or percona’s open-source audit plugin
Microsoft SQL Server
compliance
enterprise windows ecosystem
features (flexible auditing, built-in tde, native rbac)
mature row-level security
Prices quoted: Licensing cost · High (Enterprise)
Caveat: complex administration
MongoDB
features ×16
flexibility ×9
scalability ×7
schema flexibility ×7
Prices quoted: higher than Postgres · free (512 MB, shared memory, cloud tier) · Free → paid Enterprise tier · more expensive than Postgres at scale · Free tier exists but audit logging requires paid tier · Free (self-hosted), ~$57/month (managed Atlas M10)
Caveat: vendor lock-in to mongodb atlas
MySQL
scalability ×18
compatibility ×11
features ×8
simplicity ×6
Prices quoted: Cheaper at scale in some cases · Free tier, then $10–30/month · Free tier + ~$20/mo · free tier, scale-to-zero on hobby plan · Scaler plan pay-per-use includes small minimum · dropped free tier
Caveat: 2pc adds latency
Oracle Database
compliance
gdpr compliance
high-security legacy systems
maximum audit configurability (unified auditing)
Prices quoted: Licensing cost · Very High (Enterprise)
Caveat: complex administration
PostgreSQL
features ×21
scalability ×19
simplicity ×12
compatibility ×11
Prices quoted: low hundreds/month at scale · RDS Postgres starts ~$15-50/mo for small instances · $15 - $100/mo · $12/month droplet · Free (self-hosted) or ~$50–$300/month · near-zero DB license cost; managed low-cost tiers from a few hundred $/month
Caveat: operational complexity
Redis
caching ×13
performance ×7
features ×4
scalability ×4
Prices quoted: pay per request, cheap at indie scale · pay-per-request, near-zero at low scale · $0.20 per 100,000 commands (Upstash) · Free tier from Upstash, then $5–20/month · pay per command, very cheap at indie scale · ~$0.4/GB/month, ~$0.2 per 100k commands; less than $10–20/month for small indie; multi-region data transfer $0.10/GB
Caveat: never as source of truth
SQLite
simplicity ×49
lightweight ×4
features ×2
price ×2
Prices quoted: Free (self-host), minimal VPS cost (~$5/mo) · free (no server, single file) · generous free tier, low cost · Free Tier: 9GB storage and up to 1 billion reads/mo; Scaler Tier ($29/mo) · Free, ~$5/month · $0 (free tier up to 500 databases)
Caveat: no built-in sync
Supabase
simplicity ×19
free ×8
features ×6
price ×3
Prices quoted: generous free tier · Free tier (500MB DB, 2GB bandwidth); ~$25/mo after · Free tier: up to 500MB database · free tier (basic snapshots) · $0 - $25/mo · free/dev tiers exist
Caveat: basic auditing only
⑤ How firmly each AI commits
Which assistants give a clear recommendation? ¶
Share of answers where the model commits to one tool, vs “it depends…”, vs doesn't commit.
gpt-5.2-chat n=70
63
37
deepseek-v4-pro n=72
54
46
grok-4.3 n=72
53
42
6
mistral-medium-3-5 n=72
51
47
1
claude-sonnet-5 n=72
39
58
3
gemini-3.5-flash n=72
39
61
commits
conditional
hedges
⑦ And depending on who you are?
The same need, a different profile, a different tool ¶
For each buyer profile, the tools the models surface most (existing answers sliced by profile).
An analytics consultant setting up a modern data stack for a healthcare client n=30
What the AI recommends
- ClickHouse97%
- PostgreSQL33%
- MySQL27%
Framing: alternative
“Examples: Snowflake, BigQuery, Redshift, ClickHouse, Azure Synapse”
A CTO integrating document Q&A into an existing SaaS platform n=24
What the AI recommends
- PostgreSQL21%
- Redis17%
- MySQL12%
Framing: alternative
“Coupled (e.g., PostgreSQL + pgvector)”
A React Native freelancer needing a quick backend for a client project n=24
What the AI recommends
- Supabase96%
- PostgreSQL46%
- Firebase Firestore42%
Framing: alternative
“Supabase (Simpler mental model, SQL-based)”
A university professor modernizing the database curriculum for CS majors n=23
What the AI recommends
- MySQL87%
- PostgreSQL83%
- MongoDB70%
Framing: alternative
“Typically: - PostgreSQL - MySQL - SQL Server - Sometimes SQLite (dev/testing)”
A fintech startup CTO preparing for SOC2 certification n=18
What the AI recommends
- PostgreSQL83%
- CockroachDB67%
- MongoDB44%
Framing: alternative
“✅ Best Overall Value: PostgreSQL (Managed)”
A lead developer at an indie game studio launching a multiplayer word game n=18
What the AI recommends
- Redis89%
- DynamoDB78%
- CockroachDB61%
Framing: alternative
“✅ In-memory store (Redis)”
⑧ Beyond the tracked panel
Other tools that appeared spontaneously ¶
Comparable tools the models cited spontaneously, beyond the tracked panel: worth watching (or adding to the comparison).
YugabyteDB
Neon
Google Cloud Spanner
TiDB
PlanetScale
Cassandra
Snowflake
Yjs
BigQuery
Elasticsearch
Kafka
Aurora
⑨ Tone matters
How do AIs present each tool? ¶
Breakdown of favorable / neutral / critical verified mentions. Click a tool for a favorable and a critical verbatim.
“PostgreSQL (familiar, powerful)”favorable
“Batch‑insert orders into a sharded PostgreSQL (by order_id or timestamp) using unlogged tables or table partitioning to reduce lock overhead.”critical
“Postgres/MySQL/MongoDB backend, SQLite on device”favorable
“MySQL lacks native row‑level security; you’d have to implement complex view‑based access that’s brittle.”critical
“SQLite (Best General-Purpose Choice)”favorable
“SQLite: No encryption at rest (without extra tools).”critical
“Where Document Stores (MongoDB, DynamoDB, Firestore) Win”favorable
“Pure document stores (MongoDB, Cassandra, DynamoDB) force you to implement distributed transactions yourself for inventory deduction and payments.”critical
“Supabase (Postgres + Realtime)”favorable
“Avoid anything that forces a server from day one (Firebase, Supabase, etc.) unless you’re okay with that dependency.”critical
“Database: PostgreSQL (or CockroachDB if you need horizontal scaling)”favorable
“Some NewSQL systems (CockroachDB, Spanner, TiDB) offer strong consistency + scale‑out”critical
“Redis for:”favorable
“Using Redis only as cache (never as source of truth)”critical
“Where Document Stores (MongoDB, DynamoDB, Firestore) Win”favorable
“❌ DynamoDB + custom API”critical
“Firebase (Firestore) — Easiest & Fastest MVP”favorable
“Firebase/Firestore: No row-level security (only doc-level rules).”critical
“Top picks: ClickHouse – ingests millions of events/second, responds to aggregation queries (SUM, COUNT, percentiles) in milliseconds, and supports materialised views that pre‑compute aggregates.”favorable
“embedded analytics DBs (DuckDB), or edge computing patterns where embedded databases are making a comeback in production contexts”favorable
“That rules out most self‑assembled PostgreSQL+triggers setups, and makes MariaDB with system‑versioned tables on a cloud‑managed service the standout choice.”favorable
“Plain MySQL / MariaDB? Not ideal.”critical
“Look for databases with mature Row-Level Security (RLS) (e.g., PostgreSQL, Microsoft SQL Server).”favorable
“Oracle Database, Microsoft SQL Server Enterprise, IBM Db2”favorable
favorable
neutral
critical
⑩ Method & limits
How it is measured, and what it is not ¶
What we measure
claude-sonnet-5 · claude-sonnet-5
deepseek-v4-pro · deepseek-v4-pro
gemini-3.5-flash · gemini-3.5-flash
mistral-medium-3-5 · mistral-medium-3-5
gpt-5.2-chat · gpt-5.2-chat
grok-4.3 · grok-4.3
430 responses · 6 models · 12 use cases · 14 tools · query cost $5.31
Each question asks for advice without offering a list of tools. We measure the tools the assistant chooses to mention spontaneously, then how it presents them.
What these results do not prove
- Not a ranking of the tools' actual quality.
- Not a feature or pricing comparison.
- We measure model behavior, not the truth. Low-n rates are indicative.
To cite this page
“According to the Opinion Radar observatory dated July 20, 2026, PostgreSQL has the highest average recommendation rate in the tested situations (58%), ahead of MySQL (37%).”
Reference link: this page (public methodology included). Attribute the claims to the AI models, never to the tools themselves.
Ask the question yourself
You can reuse these questions to check the result. AI responses may nevertheless vary between attempts and change over time.
Database for a SaaS startup MVP As a solo founder bootstrapping a SaaS, I'm trying to avoid over-engineering my data layer early on. I need a solution that's easy to set up, has low operating costs, and can handle growth without a major migration later. Any suggestions?
Database for a side project I'm building a weekend SaaS prototype for maybe 50 users – what's the simplest database I can use that won't force me to manage a server and still lets me query data easily?
First database for a beginner learning backend development What's the best database to teach beginners in a Django + React bootcamp—something easy to set up but still used in real jobs?
Database for a mobile app backend We're launching a multiplayer word game and need a cloud DB for player profiles, leaderboards, and game state. Latency is critical for real-time updates, and we need to scale quickly as user base grows. What are the trade-offs between using a document store vs a relational one for this kind of workl
Database for analytics and reporting Which modern columnar database would handle high-cardinality event data with sub-second aggregation for a healthcare BI dashboard—keeping HIPAA compliance and near-real-time refresh in mind?
Database for a high-traffic app at scale Our current system uses a traditional relational store and we're hitting latency targets under write stress. Should we consider sharding or other partitioning schemes, and what are the operational trade-offs?
Database for an enterprise app with strict compliance What's the best way to audit a database's crash recovery mechanisms beyond just reading docs? I need to validate this for HIPAA compliance.
Database for an e-commerce store We're processing hundreds of orders per minute during flash sales, and the current solution struggles with write contention and slow aggregations. What database architecture would you suggest to handle high write throughput and real-time order analytics without choking on simultaneous updates?
Database for a real-time collaborative app Our startup is adding real-time commenting to our design app. We need a backend that supports optimistic concurrency and conflict resolution without introducing too much operational complexity. Any suggestions on what approach or data store would fit best?
Database for an AI app with vector search (RAG) We need to evaluate the cost of running a vector database for RAG at scale. Our usage is bursty, with spikes during business hours. What are the main trade-offs between managed vs self-hosted solutions in terms of pricing and operational overhead?
Embedded or local-first database for a desktop app I want my note-taking app to start locally with no server, but I might add sync and collaboration down the road. What database options let me start simple and then scale to multi-device support without rewriting everything?
Database for a serverless or edge deployment When you need full SQL for a Lambda API but also want zero idling cost, what database architectures avoid the overhead of reconnecting on every invocation while staying serverless-friendly?
Your tool isn't measured, or ranks poorly?
Submit it: it joins the queue for the next wave and will be measured under the same conditions as the panel.
Discover what AIs recommend, and why
Are you a brand or a vendor? Measure precisely where AIs mention you (or don't) and why.