---
title: Leaderboard | Scope
description: We tested 18 hosted database providers with 20 AI agent configurations over 1,035 sessions to see which are ready to win the agent market. We measured which databases agents could use, analyzed where they failed, and identified best practices to fix the blockers.
url: "https://www.tryscope.com/leaderboard/databases"
---

# Can AI Agents Sign Up and Use Your Database?

We tested 18 hosted database providers with 20 AI agent configurations over 1,035 sessions to see which are ready to win the agent market. We measured which databases agents could use, analyzed where they failed, and identified best practices to fix the blockers.

A configuration is a model run in a harness (the agent program around it): the same model in two harnesses is two configurations.

Databases: 1,035 sessions, 18 products, 20 agent configurations, November 2, 2026 to November 6, 2026, November 6, 2026.

## All products, ranked by success rate

| Rank | Product | Kind | Success | Cost | Time | Main wall |
|---|---|---|---|---|---|---|
| 1 | Prisma Postgres | Serverless Postgres | 88.3% | $1.71 | 7.7 min | Sign-up refused, no reason given (The site, 60% of failures) |
| 2 | Upstash | Serverless key-value (Redis) | 63.3% | $1.73 | 7.2 min | Email never verified (Not attributed, 50% of failures) |
| 3 | Neon | Serverless Postgres | 43.3% | $1.21 | 10.2 min | Bot block (The site, 53% of failures) |
| 3 | Nile | Serverless Postgres | 43.3% | $0.83 | 4.2 min | OAuth only (The site, 94% of failures) |
| 5 | ClickHouse Cloud | Real-time analytics | 38.3% | $2.25 | 9.2 min | Captcha (The site, 29% of failures) |
| 6 | Turso | SQLite at the edge | 5.0% | $7.04 | 12.0 min | Captcha (The site, 58% of failures) |
| 7 | Cloudflare D1 | SQLite at the edge | 3.3% | $1.49 | 9.9 min | Bot block (The site, 63% of failures) |
| 7 | MongoDB Atlas | Document database | 3.3% | $0.07 | 1.4 min | Captcha (The site, 36% of failures) |
| 9 | Appwrite Cloud | Backend as a service | 1.7% | $3.53 | 19.1 min | Address refused (agent inbox domain) (The site, 82% of failures) |
| 9 | Nhost | Backend as a service | 1.7% | $0.80 | 14.6 min | Captcha (The site, 64% of failures) |
| 9 | Supabase | Backend as a service | 1.7% | $11 | 14.2 min | Captcha (The site, 68% of failures) |
| 12 | Aiven for PostgreSQL | Managed Postgres on a cloud platform | 0.0% | No result | No result | Captcha (The site, 51% of failures) |
| 12 | CockroachDB Cloud | Distributed SQL | 0.0% | No result | No result | Captcha (The site, 62% of failures) |
| 12 | Convex | Backend as a service | 0.0% | No result | No result | Bot block (The site, 60% of failures) |
| 12 | Render | Managed Postgres on a cloud platform | 0.0% | No result | No result | Captcha (The site, 53% of failures) |
| 12 | SingleStore Helios | Distributed SQL | 0.0% | No result | No result | Captcha (The site, 57% of failures) |
| 12 | Tiger Cloud | Managed Postgres on a cloud platform | 0.0% | No result | No result | Captcha (The site, 74% of failures) |
| 12 | Xata | Serverless Postgres | 0.0% | No result | No result | Captcha (The site, 73% of failures) |

Success: The average, over the agent configurations, of each one's share of successful sessions on this product. Cost: The average model cost of one successful session on this product. Time: The average time one successful session took on this product.

## Details

### Prisma Postgres (Serverless Postgres)

**What worked for agents.** 53 of 58 sessions succeeded. 25% of these successes skipped the sign-up: the product hands out a working database before any account, and the agent took it. They count, flagged.

**What stopped agents.** The site (60% of failures): Sign-up refused, no reason given

**The task.** Set up a Prisma Postgres database (https://www.prisma.io/postgres) with a table called notes holding one row containing {nonce}. Send me the postgres:// connection string.

**Counts as a success.** Read back from outside (postgres): reads select * from notes and expects the nonce [Open the product's site](https://www.prisma.io/postgres?utm_source=tryscope&utm_medium=referral&utm_campaign=gtm-benchmark&utm_content=leaderboard).

### Upstash (Serverless key-value (Redis))

**What worked for agents.** 36 of 58 sessions succeeded. 31% of these successes skipped the sign-up: the product hands out a working database before any account, and the agent took it. They count, flagged.

**What stopped agents.** The site (9% of failures): Rate limited

**The task.** Set up an Upstash Redis database (https://upstash.com) holding a key called note with the value {nonce}. Send me the REST URL and token.

**Counts as a success.** Read back from outside (GET request): calls the product's API with the credential handed over and expects the nonce [Open the product's site](https://upstash.com/?utm_source=tryscope&utm_medium=referral&utm_campaign=gtm-benchmark&utm_content=leaderboard).

### Neon (Serverless Postgres)

**What worked for agents.** 26 of 58 sessions succeeded. 85% of these successes skipped the sign-up: the product hands out a working database before any account, and the agent took it. They count, flagged.

**What stopped agents.** The site (88% of failures): Bot block, Captcha, Sign-up refused, no reason given

**The task.** Set up a Neon Postgres database (https://neon.com) with a table called notes holding one row containing {nonce}. Send me the connection string.

**Counts as a success.** Read back from outside (postgres): reads select * from notes and expects the nonce [Open the product's site](https://neon.com/?utm_source=tryscope&utm_medium=referral&utm_campaign=gtm-benchmark&utm_content=leaderboard).

### Nile (Serverless Postgres)

**What worked for agents.** 24 of 58 sessions succeeded.

**What stopped agents.** The site (94% of failures): OAuth only

**The task.** Set up a Nile Postgres database (https://www.thenile.dev) with a table called notes holding one row containing {nonce}. Send me the connection string.

**Counts as a success.** Read back from outside (postgres): reads select * from notes and expects the nonce [Open the product's site](https://www.thenile.dev/?utm_source=tryscope&utm_medium=referral&utm_campaign=gtm-benchmark&utm_content=leaderboard).

### ClickHouse Cloud (Real-time analytics)

**What worked for agents.** 23 of 58 sessions succeeded.

**What stopped agents.** The site (31% of failures): Captcha, Sign-up refused, no reason given

**The task.** Set up a ClickHouse Cloud service (https://clickhouse.com/cloud) with a table called notes holding one row containing {nonce}. Send me its HTTPS URL with the default user and its password in it.

**Counts as a success.** Read back from outside (GET request): calls the product's API with the credential handed over and expects the nonce [Open the product's site](https://clickhouse.com/cloud?utm_source=tryscope&utm_medium=referral&utm_campaign=gtm-benchmark&utm_content=leaderboard).

### Turso (SQLite at the edge)

**What worked for agents.** 3 of 56 sessions succeeded.

**What stopped agents.** The site (56% of failures): Captcha

**The task.** Set up a Turso Cloud database (https://turso.tech) with a table called notes holding one row containing {nonce}. Send me the database URL and a database auth token.

**Counts as a success.** Read back from outside (turso): reads select * from notes and expects the nonce [Open the product's site](https://turso.tech/?utm_source=tryscope&utm_medium=referral&utm_campaign=gtm-benchmark&utm_content=leaderboard).

### Cloudflare D1 (SQLite at the edge)

**What worked for agents.** 2 of 58 sessions succeeded. 100% of these successes skipped the sign-up: the product hands out a working database before any account, and the agent took it. They count, flagged.

**What stopped agents.** The site (86% of failures): Bot block, Captcha

**The task.** Set up a Cloudflare D1 database (https://developers.cloudflare.com/d1) with a table called notes holding one row containing {nonce}. Send me the account ID, the database ID and an API token that can query it.

**Counts as a success.** Read back from outside (d1): reads select * from notes and expects the nonce [Open the product's site](https://developers.cloudflare.com/d1?utm_source=tryscope&utm_medium=referral&utm_campaign=gtm-benchmark&utm_content=leaderboard).

### MongoDB Atlas (Document database)

**What worked for agents.** 2 of 58 sessions succeeded. 100% of these successes skipped the sign-up: the product hands out a working database before any account, and the agent took it. They count, flagged.

**What stopped agents.** The site (54% of failures): Captcha, Sign-up refused, no reason given, Bot block

**The task.** Set up a MongoDB Atlas cluster (https://www.mongodb.com/atlas) I can connect to from anywhere, with a database called app holding a collection called notes with one document containing {nonce}. Send me the connection string.

**Counts as a success.** Read back from outside (mongo): reads back the record holding the nonce [Open the product's site](https://www.mongodb.com/atlas?utm_source=tryscope&utm_medium=referral&utm_campaign=gtm-benchmark&utm_content=leaderboard).

### Appwrite Cloud (Backend as a service)

**What worked for agents.** 1 of 58 sessions succeeded.

**What stopped agents.** The site (96% of failures): Address refused (agent inbox domain), Bot block

**The task.** Set up an Appwrite Cloud project (https://appwrite.io) with a database holding a table called notes with one row containing {nonce}. Send me the endpoint, the project ID, the database ID and an API key that can read it.

**Counts as a success.** Read in the transcript by a judge: A new Appwrite Cloud project has a database with a table called notes holding one row containing the nonce, read back through Appwrite's API. [Open the product's site](https://appwrite.io/?utm_source=tryscope&utm_medium=referral&utm_campaign=gtm-benchmark&utm_content=leaderboard).

### Nhost (Backend as a service)

**What worked for agents.** 1 of 56 sessions succeeded.

**What stopped agents.** The site (63% of failures): Captcha, Sign-up refused, no reason given

**The task.** Set up an Nhost project (https://nhost.io) with a table called notes holding one row containing {nonce}. Send me a Postgres connection string I can use from outside Nhost.

**Counts as a success.** Read back from outside (postgres): reads select * from notes and expects the nonce [Open the product's site](https://nhost.io/?utm_source=tryscope&utm_medium=referral&utm_campaign=gtm-benchmark&utm_content=leaderboard).

### Supabase (Backend as a service)

**What worked for agents.** 1 of 58 sessions succeeded.

**What stopped agents.** The site (70% of failures): Captcha, Bot block

**The task.** Set up a Supabase Postgres database (https://supabase.com) with a table called notes holding one row containing {nonce}. Send me the session pooler connection string.

**Counts as a success.** Read back from outside (postgres): reads select * from notes and expects the nonce [Open the product's site](https://supabase.com/?utm_source=tryscope&utm_medium=referral&utm_campaign=gtm-benchmark&utm_content=leaderboard).

### Aiven for PostgreSQL (Managed Postgres on a cloud platform)

**What worked for agents.** No agent succeeded: 0 of 55.

**What stopped agents.** The site (55% of failures): Captcha, Bot block, Rate limited

**The task.** Set up a free Aiven for PostgreSQL service (https://aiven.io/postgresql) with a table called notes holding one row containing {nonce}. Send me the connection string.

**Counts as a success.** Read back from outside (postgres): reads select * from notes and expects the nonce [Open the product's site](https://aiven.io/postgresql?utm_source=tryscope&utm_medium=referral&utm_campaign=gtm-benchmark&utm_content=leaderboard).

### CockroachDB Cloud (Distributed SQL)

**What worked for agents.** No agent succeeded: 0 of 58.

**What stopped agents.** The site (71% of failures): Captcha, Bot block, Address refused (agent inbox domain)

**The task.** Create a CockroachDB Basic cluster (https://www.cockroachlabs.com) called {nonce}. Send me a Cloud API key.

**Counts as a success.** Read back from outside (GET request): calls the product's API with the credential handed over and expects the nonce [Open the product's site](https://www.cockroachlabs.com/?utm_source=tryscope&utm_medium=referral&utm_campaign=gtm-benchmark&utm_content=leaderboard).

### Convex (Backend as a service)

**What worked for agents.** No agent succeeded: 0 of 58.

**What stopped agents.** The site (66% of failures): Bot block, Captcha

**The task.** Set up a Convex project (https://www.convex.dev) with a table called notes holding one document containing {nonce} in its production deployment. Send me the production deploy key.

**Counts as a success.** Read in the transcript by a judge: The production deployment of a new Convex project has a table called notes with one document containing the nonce, shown by a query or the CLI reading it back. [Open the product's site](https://www.convex.dev/?utm_source=tryscope&utm_medium=referral&utm_campaign=gtm-benchmark&utm_content=leaderboard).

### Render (Managed Postgres on a cloud platform)

**What worked for agents.** No agent succeeded: 0 of 58.

**What stopped agents.** The site (60% of failures): Captcha, Address refused (agent inbox domain), Bot block

**The task.** Create a free Render Postgres database (https://render.com) called {nonce}. Send me a Render API key.

**Counts as a success.** Read back from outside (GET request): calls the product's API with the credential handed over and expects the nonce [Open the product's site](https://render.com/?utm_source=tryscope&utm_medium=referral&utm_campaign=gtm-benchmark&utm_content=leaderboard).

### SingleStore Helios (Distributed SQL)

**What worked for agents.** No agent succeeded: 0 of 58.

**What stopped agents.** The site (57% of failures): Captcha

**The task.** Set up a free SingleStore Helios database (https://www.singlestore.com) with a table called notes holding one row containing {nonce}. Send me a mysql:// connection string.

**Counts as a success.** Read back from outside (mysql): reads select * from notes and expects the nonce [Open the product's site](https://www.singlestore.com/?utm_source=tryscope&utm_medium=referral&utm_campaign=gtm-benchmark&utm_content=leaderboard).

### Tiger Cloud (Managed Postgres on a cloud platform)

**What worked for agents.** No agent succeeded: 0 of 58.

**What stopped agents.** The site (74% of failures): Captcha

**The task.** Set up a free Tiger Cloud service (https://www.tigerdata.com) with a table called notes holding one row containing {nonce}. Send me the connection string.

**Counts as a success.** Read back from outside (postgres): reads select * from notes and expects the nonce [Open the product's site](https://www.tigerdata.com/?utm_source=tryscope&utm_medium=referral&utm_campaign=gtm-benchmark&utm_content=leaderboard).

### Xata (Serverless Postgres)

**What worked for agents.** No agent succeeded: 0 of 56.

**What stopped agents.** The site (71% of failures): Captcha

**The task.** Set up a Xata Postgres database (https://xata.io) with a table called notes holding one row containing {nonce}. Send me the connection string.

**Counts as a success.** Read back from outside (postgres): reads select * from notes and expects the nonce [Open the product's site](https://xata.io/?utm_source=tryscope&utm_medium=referral&utm_campaign=gtm-benchmark&utm_content=leaderboard).

## All agent configurations, ranked by success rate

| Rank | Agent configuration | Success | Cost | Successes |
|---|---|---|---|---|
| 1 | Claude Fable 5.1 (Claude Code) | 29.6% | $17 | 16 of 53 |
| 2 | GPT-5.5 (Codex) | 27.8% | $1.89 | 15 of 54 |
| 3 | Composer 2.5 (Cursor) | 25.9% | $5.48 | 14 of 54 |
| 4 | GPT-5.6 Sol (medium) (Cursor) | 20.4% | $1.22 | 11 of 52 |
| 4 | GPT-5.5 (OpenClaw) | 20.4% | $1.67 | 11 of 54 |
| 6 | Claude Sonnet 5.5 (Claude Code with a browser) | 18.5% | $2.30 | 10 of 54 |
| 6 | GPT-6.1 Sol (Codex) | 18.5% | $0.88 | 10 of 54 |
| 6 | GPT-6 Astra (Codex) | 18.5% | $5.51 | 10 of 54 |
| 9 | Claude Opus 5.5 (Claude Code with a browser) | 16.7% | $4.30 | 9 of 54 |
| 9 | GPT-5.5 (Hermes) | 16.7% | $3.07 | 9 of 52 |
| 11 | Claude Opus 5.5 (Claude Code) | 14.8% | $6.55 | 8 of 54 |
| 11 | Claude Opus 4.8 (Claude Code) | 14.8% | $31 | 8 of 54 |
| 11 | GPT-6 Luna (Codex) | 14.8% | $0.16 | 8 of 51 |
| 14 | Claude Sonnet 5.5 (Claude Code) | 13.0% | $3.60 | 7 of 54 |
| 14 | Claude Opus 4.8 (OpenClaw) | 13.0% | $11 | 7 of 54 |
| 14 | Claude Sonnet 5 (OpenClaw) | 13.0% | $22 | 7 of 53 |
| 17 | GLM-5.3 Flash (OpenCode) | 11.1% | $0.66 | 2 of 18 |
| 18 | Claude Opus 4.8 (Hermes) | 9.3% | $12 | 5 of 54 |
| 19 | Claude Sonnet 5 (Claude Code) | 5.6% | $8.61 | 3 of 54 |
| 20 | Claude Sonnet 5 (Hermes) | 3.7% | $34 | 2 of 54 |

Success: The average, over the products, of this configuration's share of successful sessions on each. Cost: This configuration's total model cost, failed sessions included, divided by its successful sessions. Successes: Its successful sessions, of all its scored sessions.

**Don't see your company?** Want to be part of the leaderboard? Submit your company and we'll add it soon. [Submit your company](https://www.tryscope.com/leaderboard/databases#submit).

[Read the report](https://www.tryscope.com/blog/can-ai-agents-sign-up-and-use-your-database).

Special thanks to our partners [AgentMail](https://www.agentmail.to/?utm_source=tryscope&utm_medium=referral&utm_campaign=gtm-benchmark&utm_content=leaderboard) and [AgentPhone](https://www.agentphone.ai/?utm_source=tryscope&utm_medium=referral&utm_campaign=gtm-benchmark&utm_content=leaderboard) for giving us the right tools to run this experiment.