MultiVerse_70B: benchmark results
MTS AI's chat model built from Qwen 72B with an unpublished proprietary 'multiverse' training method, despite the 70B name. Provider: MTS AI. Released 2024-03-25. Access: Open.
Aggregate Rating 1642 ± 43, rank #678 of 10916 rated models, from 14 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open Chinese LLM - C-Eval Semantic | 91.53 | Accuracy (%) | 99.4 |
| Open Chinese LLM - ARC Challenge | 69.28 | Accuracy (%) | 98.8 |
| Open Chinese LLM - HellaSwag | 77.08 | Accuracy (%) | 98.5 |
| Open Chinese LLM Leaderboard | 72.08 | Official Nonzero-Task Mean (%) | 98.2 |
| Open Chinese LLM - WinoGrande | 71.74 | Accuracy (%) | 97.6 |
| Open Chinese LLM - TruthfulQA MC | 66.1 | Accuracy (%) | 97 |
| Open Chinese LLM - GSM8K | 69.29 | Accuracy (%) | 94.2 |
| Open LLM Leaderboard - MuSR | 47.4 | Score | 92.7 |
| Open Chinese LLM - CMMLU | 71.61 | Accuracy (%) | 92.5 |
| Open LLM Leaderboard - GPQA | 35.4 | Score | 89.1 |
| Open LLM Leaderboard - MMLU-Pro | 48.6 | Score | 86.8 |
| Open LLM Leaderboard - BBH | 61.83 | Score | 85.5 |
Interactive version: theaggregate.ai/model?slug=multiverse-70b · How It Works · Data refreshed daily, snapshot 2026-11-08.