Buttercup-4x7B-bf16: benchmark results
Kquant03's mixture-of-experts release whose published Mixtral configuration specifies 4 local experts, 2 selected per token and 32 decoder layers. Its recipe names mlabonne/Beagle14-7B as the base; nominal expert sizes are not measured total or active parameter counts. Provider: Kquant03. Access: Open.
Aggregate Rating 1456 ± 39, rank #3511 of 10916 rated models, from 6 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open LLM Leaderboard v1 - ARC Challenge | 72.1 | Normalized accuracy (%) (25-shot) | 91 |
| Open LLM Leaderboard v1 - GSM8K | 69.83 | Accuracy (%) (5-shot) | 90.6 |
| Open LLM Leaderboard v1 - HellaSwag | 87.74 | Normalized accuracy (%) (10-shot) | 87.6 |
| Open LLM Leaderboard v1 - TruthfulQA MC2 | 67.2 | MC2 (%) (0-shot) | 85.6 |
| Open LLM Leaderboard v1 - WinoGrande | 81.93 | Accuracy (%) (5-shot) | 81.3 |
| Open LLM Leaderboard v1 - MMLU | 64.58 | Accuracy (%) (5-shot) | 72.9 |
Interactive version: theaggregate.ai/model?slug=buttercup-4x7b-bf16 · How It Works · Data refreshed daily, snapshot 2026-11-08.