Llama 3 8B ProLong 512k Instruct Benchmark Results

Llama 3 8B ProLong 512k Instruct — benchmark results

Princeton NLP's ProLong 512K, extending Llama 3 8B Instruct to a 512K context via 40B tokens of continued training plus UltraChat SFT. Provider: Meta. Released 2024-08-22. Access: Open.

Unified ELO 1404 ± 10, rank #1205 of 1776 rated models, from 15 benchmark results.

Strongest benchmark results

Benchmark	Score	Metric	Percentile
Open Korean LLM Leaderboard	451.44	Average Score (%)	82.8
Open LLM Leaderboard - MuSR	12.53	Score	68.6
Open LLM Leaderboard - IFEval	55.08	Score	66.5
Open LLM Leaderboard - BBH	29.15	Score	48.7
Open Chinese LLM - TruthfulQA MC	53.45	Accuracy (%)	48.1
Open LLM Leaderboard - MMLU-Pro	24.96	Score	44.7
Open Chinese LLM - ARC Challenge	50.09	Accuracy (%)	41.4
Open LLM Leaderboard - GPQA	4.81	Score	41
Open Chinese LLM - CMMLU	51.79	Accuracy (%)	40.4
Open Chinese LLM - GSM8K	40.86	Accuracy (%)	40.2
Open Chinese LLM - C-Eval Semantic	64.65	Accuracy (%)	37.7
Open Chinese LLM - HellaSwag	57.65	Accuracy (%)	33.8

Interactive version: theaggregate.ai/model?slug=llama-3-8b-prolong-512k-instruct · How the rankings work · Data refreshed daily, snapshot 2026-07-22.