FuseChat-Llama-3.2-3B-Instruct — benchmark results

FuseAI's implicit-model-fusion tune of Llama 3.2 3B Instruct, distilling preferences from Gemma 2 27B, Mistral Large, Qwen2.5 72B, and Llama 3.1 70B. Provider: Other. Released 2024-12-06. Access: Open.

Unified ELO 1514 ± 46, rank #755 of 1776 rated models, from 7 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
AlpacaEval 2.054LC Win Rate (%)90.1
Open LLM Leaderboard - IFEval68.49Score81.2
Open LLM Leaderboard - MATH Level 524.24Score77.5
Open LLM Leaderboard - GPQA6.15Score51.6
Open LLM Leaderboard - MMLU-Pro23.68Score41.3
Open LLM Leaderboard - BBH24.22Score37
Open LLM Leaderboard - MuSR7.69Score35.7

Interactive version: theaggregate.ai/model?slug=fusechat-llama-3-2-3b-instruct · How the rankings work · Data refreshed daily, snapshot 2026-07-22.