Orca-2-13B: benchmark results
Microsoft's Orca 2 13B, a research-only model built on the Llama 2 13B base for single-turn reasoning, reading comprehension, math and summarization tasks, not yet tuned for chat. Provider: Microsoft. Released 2023-11-14. Access: Open.
Aggregate Rating 1316 ± 34, rank #1448 of 1656 rated models, from 18 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open LLM Leaderboard - MuSR | 51.3 | Score | 99.2 |
| Hallucinations Leaderboard | 44.31 | Average Task Score (score) | 55.3 |
| Open LLM Leaderboard - BBH | 48.84 | Score | 43.7 |
| pfgen-bench - Completion Mode - Fluency | 0.51 | Fluency Score | 38 |
| Open CoT - LSAT Logical Reasoning | 7.84 | CoT Gain (%) | 37 |
| Open CoT - LogiQA 2 | 4.58 | CoT Gain (%) | 36.3 |
| pfgen-bench - Completion Mode - Helpfulness | 0.07 | Helpfulness Score | 36.1 |
| Open LLM Leaderboard - GPQA | 28.02 | Score | 35.2 |
| pfgen-bench - Completion Mode - Score | 0.4 | pfgen Score (mean of three) | 34.5 |
| pfgen-bench - Completion Mode - Truthfulness | 0.63 | Truthfulness Score | 31.6 |
| Open LLM Leaderboard - MMLU-Pro | 27.49 | Score | 29.9 |
| Open LLM Leaderboard - IFEval | 31.28 | Score | 28.9 |
Interactive version: theaggregate.ai/model?slug=orca-2-13b · How It Works · Data refreshed daily, snapshot 2026-10-20.