spring-chicken-8x8B: benchmark results

maldv Spring Chicken is an eight-expert Llama 3 mixture with two experts selected per token and hidden-state gating. Its Spring base is a cascading DARE-TIES merge, distinct from the four-expert Badger mixtures. Provider: maldv. Access: Open.

Aggregate Rating 1413 ± 33, rank #5514 of 10916 rated models, from 6 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
Open LLM Leaderboard v1 - MMLU64.45Accuracy (%) (5-shot)70.7
Open LLM Leaderboard v1 - GSM8K57.47Accuracy (%) (5-shot)69.2
Open LLM Leaderboard v1 - ARC Challenge63.05Normalized accuracy (%) (25-shot)61
Open LLM Leaderboard v1 - TruthfulQA MC251.63MC2 (%) (0-shot)52.6
Open LLM Leaderboard v1 - HellaSwag82.49Normalized accuracy (%) (10-shot)51.6
Open LLM Leaderboard v1 - WinoGrande76.24Accuracy (%) (5-shot)47

Interactive version: theaggregate.ai/model?slug=spring-chicken-8x8b · How It Works · Data refreshed daily, snapshot 2026-11-08.