Nanbeige4.3 Base Release - CMMLU: leaderboard
CMMLU Chinese multiple-choice knowledge questions. Nanbeige4.3-33B-Base model card (Hugging Face, 2026-10-25): Nanbeige compared its pretrained base model with Gemma4-31B-Base. These are base checkpoints, not the instruction-tuned Nanbeige4.3-33B release; shot counts, prompts, decoding and the exact benchmark variant are not stated. Score (%), higher is better.
Metric: Score (%). Source: huggingface.co. Saturation forecast: Rough model projection: around 2026. 2 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Nanbeige4.3-33B-Base | 86.8 | |
| 2 | Gemma4-31B-Base | 76.9 |
Interactive version: theaggregate.ai/benchmark?slug=nanbeige4-3-base-release-cmmlu · How It Works · Data refreshed daily, snapshot 2026-11-05.