RABBITS: leaderboard
Bias assessment for Indian text-based systems: evaluates medical LLMs for bias and accuracy on MedMCQA and MedQA with culturally-adapted questions.
Metric: B4BQA Score (%). Source: huggingface.co. Status: saturated. 23 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-4 | 99.71 |
| 2 | GPT-4o | 99.61 |
| 3 | GPT-3.5 Turbo | 99.43 |
| 4 | Claude Opus | 99.14 |
| 5 | Qwen-Qwen2-72B | 99.05 |
| 6 | mistralai-Mixtral-8x22B-v0.1 | 98.66 |
| 7 | meta-llama-Meta-Llama-3-70B | 98.57 |
| 8 | CohereForAI-c4ai-command-r-plus | 98.28 |
| 9 | 01-ai-Yi-1.5-34B | 97.81 |
| 10 | aaditya-Llama3-OpenBioLLM-70B | 97.42 |
Interactive version: theaggregate.ai/benchmark?slug=rabbits · How It Works · Data refreshed daily, snapshot 2026-09-05.