gstack-benchmark-models

What problem does it solve?

This Skill removes guesswork when choosing which AI model is best for a given task by running the same prompt across providers and comparing measurable outcomes like latency and cost.

Core Features & Use Cases

• Cross-model prompt shootouts: Executes the same prompt through Claude, GPT (via Codex CLI), and Gemini side-by-side to compare performance.
• Actionable comparisons: Reports latency, token usage, and cost, and can optionally add an LLM judge score for output quality.
• Safe preflight workflow: Forces an auth-aware dry-run first so you can see what providers are ready before any paid benchmark run.
• Repeatable baselines: Can save benchmark results as JSON so you can compare future runs and detect performance drift.

Quick Start

Run gstack-benchmark-models and choose a prompt (or select a benchmark-able gstack skill) to compare Claude, GPT, and Gemini on the same input.