About Banzuke
Mission
Banzuke is a consensus-ranking platform that aggregates independent votes from multiple AI models, reducing reliance on any single model's bias. While any individual AI may produce biased or inaccurate output, results that multiple AIs agree on tend to be more reliable. We operate this platform with transparency of the evaluation process as our top priority.
Methodology
Our editorial team submits the same prompt and evaluation axes to three AI models — Claude (Anthropic), GPT (OpenAI), and Gemini (Google) — and asks each to independently select its Top 10. The results are aggregated using the Borda Count method (1st place = 10 points, 10th = 1 point) to determine the final ranking. Each AI's reasoning, evaluation axes, and scores are fully disclosed on the banzuke detail page.
AI Models Used
We currently use the following AI models. • Claude (Anthropic, PBC) • GPT (OpenAI, L.P.) • Gemini (Google LLC) Each model's name and trademark belong to their respective providers. Banzuke uses these names solely for identification purposes; this does not imply endorsement or partnership.
API & Data Access
We do not currently offer a public API. For inquiries about data licensing for businesses or research institutions, please contact us at the address below.
Operating Company
| Company name | Resolve Inc. |
|---|---|
| Representative | Hiroshi Nakajima |
| Address | Shibuya Dogenzaka Tokyu Bldg. 2F-C, 1-10-8 Dogenzaka, Shibuya-ku, Tokyo, Japan |
| Contact | support@banzuke.ai |