OpenGPT

OpenGPT provides an AI decision workspace that helps users compare candidates for AI models and configurations. It emphasizes explainability by separating public facts, editorial judgment, user conditions, and browser-local evaluation. The system operates in a way that aims to keep a shortlist explainable rather than presenting a single verdict.

Builder turns a local Profile into a deterministic, conditional shortlist. Workspace supports exact-version comparison and representative work tests. Cost intelligence is planning guidance, not a bill or price guarantee. Decision intelligence separates sourced facts from OpenGPT editorial judgment.

Workspace is designed so the user can compare exact versions and test candidates with representative work, keeping local prompts, pasted answers, judgments, and reports in the browser unless explicitly exported.

OpenGPT does not claim that one ranking identifies the best system for every person or task and does not sell ranking placement, confidence labels, or editorial outcomes. It also does not silently upload local prompts, answers, or reports.

Use cases

• Comparison and evaluation of AI models and configurations for tasks
• Cost-informed planning for model choices