Open source tool measures the stupidity level of AI models

[Redaktion]

A new open-source tool is offering real-time monitoring of multiple AI models, including OpenAI GPT-5, Claude Opus 4, and Gemini 2.5 Pro. The first of its kind, it can detect "when AI companies reduce model capability to save costs." The benchmarks can run against the users' own OpenAI, xAI, Anthropic, or Google API keys as well.

https://www.notebookcheck.net/Open-source-tool-measures-the-stupidity-level-of-AI-models.1117315.0.html