MiniCPM5-2B: benchmark results
A dense LlamaForCausalLM checkpoint with 2,516,756,480 total parameters, including 1,981,982,720 non-embedding parameters, and native 131,072-token context. This is the final post-trained release rather than a base or SFT-only checkpoint. Provider: OpenBMB. Released 2026-09-07. Access: Open.
Aggregate Rating 1558 ± 20, rank #1458 of 10524 rated models, from 85 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| LFM3-1.7B Release - τ³-bench (4 domains), average | 69.44 | Score (%) | 100 |
| AA-Omniscience Hallucination Rate | 21.91 | Hallucination Rate (%) | 97.8 |
| LFM3-1.7B Release - AA-Omniscience (public) | -5 | Omniscience net score | 80 |
| LFM3-1.7B Release - AA-Omniscience non-hallucination (public) | 84.44 | Score (%) | 80 |
| LFM3-1.7B Release - IFBench | 62.08 | Score (%) | 80 |
| AA Omniscience | -11.65 | Score | 62 |
| Docbench - Planted Problem Recall - Substantial | 13 | Planted problems found (%) | 61.1 |
| LFM3-1.7B Release - BFCL V4 | 62.65 | Score (%) | 60 |
| LFM3-1.7B Release - GPQA Diamond | 65.15 | Score (%) | 60 |
| LFM3-1.7B Release - Humanity's Last Exam (no tools) | 8.2 | Score (%) | 60 |
| LFM3-1.7B Release - LiveClaw | 30.37 | Score (%) | 60 |
| LFM3-1.7B Release - LiveCodeBench v6 | 70.52 | Score (%) | 60 |
Interactive version: theaggregate.ai/model?slug=minicpm5-2b · How It Works · Data refreshed daily, snapshot 2026-11-22.