internlm2.5-7B: benchmark results
Shanghai AI Lab's 7B InternLM2.5 base model with ~20% better reasoning than InternLM2 and support for up to 1M-token context (mid-2024). Provider: Shanghai AI Lab. Released 2024-07-03. Access: Open.
Aggregate Rating 1423 ± 22, rank #1133 of 1656 rated models, from 42 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| GSMA Open-Telco - TeleTables | 30.13 | Score (%) | 57.3 |
| Open PL LLM - PPC (multiple choice, 5-shot) | 70.9 | Accuracy (%) | 54.2 |
| Open PL LLM - PSC (multiple choice, 5-shot) | 66.04 | Binary F1 (%) | 53.3 |
| Open PL LLM - KLEJ NER (multiple choice, 5-shot) | 47.04 | Accuracy (%) | 53.1 |
| Open PL LLM - PolEmo2-IN (generative, 5-shot) | 75.76 | Accuracy (%) | 51.4 |
| Open PL LLM - PoQuAD Open Book (generative, 5-shot) | 62.04 | Levenshtein Similarity (%) | 50.2 |
| Open PL LLM - PSC (generative, 5-shot) | 87.87 | Binary F1 (%) | 49.8 |
| Open PL LLM - KLEJ NER (generative, 5-shot) | 44.27 | Accuracy (%) | 48 |
| Open PL LLM - Belebele (generative, 5-shot) | 74.11 | Accuracy (%) | 45.3 |
| Open PL LLM - Belebele (multiple choice, 5-shot) | 74.78 | Accuracy (%) | 45 |
| Open PL LLM - PolEmo2-OUT (generative, 5-shot) | 65.59 | Accuracy (%) | 44.8 |
| Open PL LLM - Generative | 44.4 | Average Generative Score (%) | 43 |
Interactive version: theaggregate.ai/model?slug=internlm2-5-7b · How It Works · Data refreshed daily, snapshot 2026-10-20.