OpenGPTEvidence-based model decisionsABOUT OPENGPT
Evidence before confident model decisions
OpenGPT is an independent, provider-neutral guide to public AI model and Agent evidence. It helps people move from discovery to a shortlist, a fair comparison, and a decision with visible limits.
Last updated August 3, 2026
- Public rankings keep their original source and scope.
- Model and Agent identities are separated from benchmark claims.
- Real-task evaluation stays in the browser unless a user explicitly exports it.
- No provider, sponsor, or submitter can buy ranking position.
What OpenGPT does
OpenGPT organizes public source records, provider-documented model IDs, Agent product references, and comparable benchmark configurations. It also provides local tools for shortlisting, side-by-side review, real-task evaluation, and report verification.
What OpenGPT does not do
OpenGPT does not claim that one ranking identifies the best system for every person or task. Unless a result explicitly says OpenGPT reproduced it, the site is organizing upstream evidence rather than rerunning the evaluation.
- It does not sell ranking placement or trust grades.
- It does not silently upload local prompts, answers, or reports.
- It is not affiliated with or endorsed by OpenAI or any other model provider.
How to use the site
Start with a task, inspect two or three candidates, read the evidence boundary, and test exact versions with representative work. Missing data remains unknown instead of being converted to zero or an estimate.
Contact
Report an inaccurate identity mapping, stale official link, unsupported claim, or broken page at contact@opengpt.com.