# JudgeBench evaluated system AI models

JudgeBench evaluated system has 2 tracked model families. Arena-Hard (Mistral-7B-v0.1-Instruct) (JudgeBench) is its highest current public-ranked model.

| Model | Public score | Confidence | Context | Input / output price per 1M |
|---|---:|---|---:|---:|
| [Arena-Hard (Mistral-7B-v0.1-Instruct) (JudgeBench)](https://benchlm.ai/models/native-judgebench-arena-hard-mistral-7b-v0-1-instruct) | Not ranked | Estimated | N/A | Pricing unavailable / Pricing unavailable |
| [Vanilla (Mistral-7B-v0.1-Instruct) (JudgeBench)](https://benchlm.ai/models/native-judgebench-vanilla-mistral-7b-v0-1-instruct) | Not ranked | Estimated | N/A | Pricing unavailable / Pricing unavailable |

Last updated: October 1, 2026

Canonical page: https://benchlm.ai/providers/judgebench-evaluated-system