MixTAO-7Bx2-MoE-v8.1: benchmark results
zhengr's experimental 2x7B Mistral mixture-of-experts (~13B), a top scorer among small MoEs on the early-2024 Open LLM Leaderboard. Provider: mixtao. Released 2024-02-26. Access: Open.
Aggregate Rating 1485 ± 17, rank #827 of 1656 rated models, from 27 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open LLM Leaderboard v1 - WinoGrande | 87.37 | Accuracy (%) (5-shot) | 99.6 |
| Open LLM Leaderboard v1 - TruthfulQA MC2 | 78.57 | MC2 (%) (0-shot) | 99.3 |
| Open LLM Leaderboard v1 - HellaSwag | 89.22 | Normalized accuracy (%) (10-shot) | 98.6 |
| Open LLM Leaderboard v1 - ARC Challenge | 73.81 | Normalized accuracy (%) (25-shot) | 98.3 |
| Open Korean LLM - KorNAT-Harmless | 77.91 | Normalized accuracy (%) | 98.2 |
| Open Korean LLM - KorNAT-CKA | 35.89 | Normalized accuracy (%) | 95.9 |
| Open LLM Leaderboard v1 - GSM8K | 71.11 | Accuracy (%) (5-shot) | 95.2 |
| Open Korean LLM - KorNAT-Helpful | 52.1 | Normalized accuracy (%) | 92.1 |
| Open Korean LLM Leaderboard | 44.46 | Average Score (%) | 90.3 |
| Open Ko-LLM Leaderboard | 44.46 | Average (%) | 90.2 |
| Open Korean LLM - Ko-Winogrande | 65.75 | Accuracy (%) | 86 |
| Open LLM Leaderboard - MuSR | 44.63 | Score | 81.6 |
Interactive version: theaggregate.ai/model?slug=mixtao-7bx2-moe-v8-1 · How It Works · Data refreshed daily, snapshot 2026-10-20.