granite-3.0-8B-instruct — benchmark results

IBM's Apache-2.0 Granite 3.0 8B instruct model (October 2024), an enterprise workhorse tuned for RAG, summarization, extraction, and tool use. Provider: IBM. Released 2024-10-21. Access: Open.

Unified ELO 1459 ± 11, rank #960 of 1776 rated models, from 54 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
Open LLM Leaderboard - GPQA10.96Score81.4
EuroEval Spanish NLU - Sentiment Headlines ES48.18Sentiment classification Score (%)81.3
EuroEval Spanish NLU - MLQA ES64.53Reading comprehension Score (%)80.6
EuroEval Swedish NLU - Swerec77.96Sentiment classification Score (%)80.6
EuroEval Portuguese NLU - MultiWikiQA PT72.15Reading comprehension Score (%)68.4
EuroEval Italian NLU - SQuAD IT69.69Reading comprehension Score (%)66.5
EuroEval Danish NLU - Angry Tweets50.78Sentiment classification Score (%)63.9
Open LLM Leaderboard - IFEval53.1Score63.1
Open LLM Leaderboard - MATH Level 514.2Score60.6
EuroEval Spanish NLU45.57NLU Average Score (%)59.8
EuroEval Portuguese NLU - SST-2 PT78.64Sentiment classification Score (%)58.9
Open LLM Leaderboard - BBH31.59Score57.6

Interactive version: theaggregate.ai/model?slug=granite-3-0-8b-instruct · How the rankings work · Data refreshed daily, snapshot 2026-07-22.