Compare AI Setups From Your Terminal

Ask Claude Code or Codex to run the same task with different models or skills. Start in the terminal and review the outputs in Evals.

Use the CLIExplore Public Outputs

Start a Comparison

  1. Run evals login and approve your terminal once.
  2. In Claude Code, enter /evals and describe what to compare. In Codex, use $evals.
  3. Check the setups, visibility, and payment routes. Confirm the run in your assistant or terminal.
Manage CLI Devices

Compare a Skill Against the Baseline

Give both setups the same model and prompt. Select a skill for one setup and choose no skills for the other.

Find a Skill

Review What Each Setup Made

Open the link returned by the CLI to inspect the outputs, recorded runtime, and usage in Evals.

How Evals Compares