OLMo-7B-0724-Instruct-hf Benchmark Results

OLMo-7B-0724-Instruct-hf: benchmark results

Provider: Allen AI. Access: Open.

Unified ELO 1305 ± 25, rank #1310 of 1516 rated models, from 103 benchmark results.

Benchmark	Score	Metric	Percentile
AbstentionBench - underspecified context - QASPER - F1	45.48	Abstention F1 (%)	100
AbstentionBench - underspecified context - QASPER - Precision	30.17	Abstention Precision (%)	100
AbstentionBench - stale - FreshQA - Recall	89.36	Abstention Recall (%)	97.4
Open CoT - LSAT Reading Comprehension	18.59	CoT Gain (%)	80.2
AbstentionBench - answer unknown - BB/Known unknowns - Recall	100	Abstention Recall (%)	71.1
AbstentionBench - answer unknown - CoCoNot/Unsupported - Precision	98.78	Abstention Precision (%)	63.2
AbstentionBench - false premise - FalseQA - Recall	55.17	Abstention Recall (%)	63.2
AbstentionBench - false premise - QAQA - Recall	48.77	Abstention Recall (%)	63.2
AbstentionBench - stale - FreshQA - F1	47.19	Abstention F1 (%)	63.2
AbstentionBench - false premise - FalseQA - F1	65.12	Abstention F1 (%)	57.9
AbstentionBench - underspecified context - ALCUNA - Recall	70.73	Abstention Recall (%)	57.9
AbstentionBench - underspecified context - BB/Disambiguate - Precision	52.38	Abstention Precision (%)	57.9

Interactive version: theaggregate.ai/model?slug=olmo-7b-0724-instruct-hf · How It Works · Data refreshed daily, snapshot 2026-09-24.