InternLM2.5-20B: benchmark results
InternLM team's internlm2_5-20b base release. Its card describes retaining the InternLM2 architecture while using synthetic data in iterative model development, with an emphasis on reasoning. Provider: Shanghai AI Lab. Access: Open.
Aggregate Rating 1429 ± 50, rank #4720 of 10916 rated models, from 13 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open Portuguese LLM - FaQuAD NLI | 73.02 | Macro F1 (%) | 67.5 |
| Open Portuguese LLM - ASSIN2 RTE | 91.22 | Macro F1 (%) | 65.1 |
| GSMA Open-Telco - TeleLogs | 15.66 | Score (%) | 33.7 |
| GSMA Open-Telco - TeleQnA | 67.21 | Score (%) | 24.7 |
| GSMA Open-Telco - 3GPP | 30.65 | Score (%) | 23.6 |
| GSMA Open-Telco - TeleTables | 25.6 | Score (%) | 16.9 |
| GSMA Open-Telco LLM Leaderboard | 40.59 | Average Score (%) | 15.7 |
| GSMA Open-Telco - TeleMath | 12.6 | Score (%) | 15.2 |
| GSMA Open-Telco - srsRAN-Bench | 70.11 | Score (%) | 13.5 |
| GSMA Open-Telco - ORAN-Bench | 62.31 | Score (%) | 11.2 |
| Open Portuguese LLM - ENEM | 2.73 | Accuracy (%) | 2.7 |
| Open Portuguese LLM - OAB Exams | 1.18 | Accuracy (%) | 2.6 |
Interactive version: theaggregate.ai/model?slug=internlm2-5-20b · How It Works · Data refreshed daily, snapshot 2026-11-08.