Get early access.

Run dozens of models and their harnesses (Claude Code, Codex, Cursor, etc.) against your docs, SDK and MCP server, to see which tasks they complete and where they fail.

Tell us about yourself and we’ll add you to the first cohort.

ALPIC

EXPERIENCE

Experiments

Runs

Library

Analytics

Experiments / Auth SDK: Next.js quickstart

Auth SDK: Next.js quickstart

Relaunch

Execution #2 · v2

Ended

40%

2 / 5 runs

$2.41

total spend

Scenario

Claude Code

Codex

Cursor

Add sign-in from the docs

0%

0/1 runs

100%

1/1 runs

0%

0/1 runs

Create a project via MCP

100%

1/1 runs

100%

1/1 runs

0%

0/1 runs

New run

2 agents × 3 runs = 6 runs

Start run

Prompt

Add passwordless sign-in to this Next.js app with our auth SDK. Use only the public docs and the MCP server.

Agents

2 selected

Claude Code
Codex
Cursor
Gemini CLI

Runs per agent

3
−
+

MCP server

auth-docs (MCP)⌄

Budget per run

$ 1

Time limit (min)

10

Max spend: $6