zephyr-orpo-141B-A35B-v0.1: benchmark results
Hugging Face H4's ORPO preference-tuned Mixtral 8x22B (141B total), built with Argilla and KAIST. Provider: HuggingFace. Released 2024-04-10. Access: Open.
Aggregate Rating 1536 ± 51, rank #1871 of 10916 rated models, from 29 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open LLM Leaderboard - GPQA | 37.84 | Score | 94.3 |
| BenchBench | 84.14 | Aggregate Score (%) | 91.2 |
| Open LLM Leaderboard - BBH | 62.9 | Score | 86.6 |
| Open LLM Leaderboard - MMLU-Pro | 45.86 | Score | 84.7 |
| Open LLM Leaderboard - MuSR | 44.65 | Score | 81.7 |
| Open LLM Leaderboard - IFEval | 65.11 | Score | 77.4 |
| Open LLM Leaderboard - MATH Level 5 | 20.47 | Score | 71.8 |
| Chatbot Arena (Text - Entertainment, Sports & Media) | 1190 | Arena Score | 20.4 |
| Chatbot Arena (Text - English) | 1240 | Arena Score | 18.9 |
| Chatbot Arena (Text - Life, Physical & Social Science) | 1238 | Arena Score | 18.5 |
| Chatbot Arena (Text - Instruction Following) | 1194 | Arena Score | 18.4 |
| Chatbot Arena (Text - Writing, Literature & Language) | 1196 | Arena Score | 18.4 |
Interactive version: theaggregate.ai/model?slug=zephyr-orpo-141b-a35b-v0-1 · How It Works · Data refreshed daily, snapshot 2026-11-08.