Llama 3.1 8B Cobalt Benchmark Results

Llama 3.1 8B Cobalt — benchmark results

Valiant Labs' math-instruct fine-tune of Llama 3.1 8B Instruct, trained on synthetic math data generated with Llama 3.1 405B. Provider: Meta. Released 2024-08-16. Access: Open.

Unified ELO 1467 ± 34, rank #925 of 1776 rated models, from 7 benchmark results.

Strongest benchmark results

Benchmark	Score	Metric	Percentile
Open LLM Leaderboard - IFEval	71.68	Score	85.2
Open Korean LLM Leaderboard	368.64	Average Score (%)	73.4
Open LLM Leaderboard - MATH Level 5	15.33	Score	62.5
Open LLM Leaderboard - GPQA	7.16	Score	60.3
Open LLM Leaderboard - MMLU-Pro	29.59	Score	57.6
Open LLM Leaderboard - MuSR	9.83	Score	48.4
Open LLM Leaderboard - BBH	27.42	Score	43.7

Interactive version: theaggregate.ai/model?slug=llama-3-1-8b-cobalt · How the rankings work · Data refreshed daily, snapshot 2026-07-22.