34B-beta: benchmark results
CausalLM's beta 34B ChatML chat model initialized from Yi-34B, released with noted weight-precision caveats for accelerated inference. Provider: CausalLM. Released 2024-02-06. Access: Open.
Aggregate Rating 1466 ± 23, rank #903 of 1656 rated models, from 25 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open LLM Leaderboard v1 - MMLU | 85.6 | Accuracy (%) (5-shot) | 99.9 |
| Open Chinese LLM - CMMLU | 71.71 | Accuracy (%) | 93.5 |
| Open LLM Leaderboard - MMLU-Pro | 53.25 | Score | 93.3 |
| Open Portuguese LLM - BLUEX | 66.06 | Accuracy (%) | 87.8 |
| Open LLM Leaderboard - GPQA | 34.65 | Score | 86.7 |
| Open LLM Leaderboard v1 - ARC Challenge | 70.56 | Normalized accuracy (%) (25-shot) | 86.4 |
| Open Chinese LLM - C-Eval Semantic | 87.63 | Accuracy (%) | 85.8 |
| Open Portuguese LLM - OAB Exams | 53.49 | Accuracy (%) | 83.7 |
| Open Portuguese LLM - ENEM | 72.43 | Accuracy (%) | 83.2 |
| Open Chinese LLM - ARC Challenge | 59.22 | Accuracy (%) | 82.8 |
| Open Chinese LLM - HellaSwag | 66.42 | Accuracy (%) | 81.2 |
| Open Chinese LLM Leaderboard | 64.7 | Official Nonzero-Task Mean (%) | 80.8 |
Interactive version: theaggregate.ai/model?slug=34b-beta · How It Works · Data refreshed daily, snapshot 2026-10-20.