mera-mix-4x7B: benchmark results
meraGPT's 4-expert Mixtral-style MoE pitched as a half-size, faster drop-in replacement for Mixtral-8x7B at comparable benchmark scores. Provider: meraGPT. Released 2024-04-13. Access: Open.
Aggregate Rating 1463 ± 23, rank #919 of 1656 rated models, from 25 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open LLM Leaderboard v1 - WinoGrande | 85.64 | Accuracy (%) (5-shot) | 98.7 |
| Open LLM Leaderboard v1 - HellaSwag | 89.17 | Normalized accuracy (%) (10-shot) | 98.2 |
| Open LLM Leaderboard v1 - TruthfulQA MC2 | 77.17 | MC2 (%) (0-shot) | 96.6 |
| Open LLM Leaderboard v1 - ARC Challenge | 72.95 | Normalized accuracy (%) (25-shot) | 94.5 |
| Open Chinese LLM - TruthfulQA MC | 62.51 | Accuracy (%) | 90.5 |
| Open LLM Leaderboard v1 - GSM8K | 66.11 | Accuracy (%) (5-shot) | 81.2 |
| Open Portuguese LLM - FaQuAD NLI | 77.33 | Macro F1 (%) | 79.4 |
| Open Portuguese LLM - ASSIN2 RTE | 92.16 | Macro F1 (%) | 75.3 |
| Open LLM Leaderboard v1 - MMLU | 64.44 | Accuracy (%) (5-shot) | 70.4 |
| Open Chinese LLM - WinoGrande | 64.01 | Accuracy (%) | 63.2 |
| Open Portuguese LLM - BLUEX | 55.08 | Accuracy (%) | 62.2 |
| Open LLM Leaderboard - GPQA | 30.45 | Score | 61.3 |
Interactive version: theaggregate.ai/model?slug=mera-mix-4x7b · How It Works · Data refreshed daily, snapshot 2026-10-20.