Llama 2 13B Chat: benchmark results
Meta's Llama 2 13B Chat, the RLHF fine-tuned chat version of Llama 2 13B, released under the Llama 2 Community License in July 2023; the linked card provides no further details here. Provider: Meta. Released 2023-07-18. Access: Open.
Aggregate Rating 1307 ± 11, rank #1465 of 1656 rated models, from 202 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| CLAMBER - Zero-shot - Identification Accuracy | 55.47 | Accuracy (%) | 100 |
| MT-Bench | 6.65 | MT-Bench average score (points) | 78.8 |
| CLAMBER - Few-shot CoT - Clarification BERTScore | 23.68 | BERTScore (0-100) | 75 |
| CLAMBER - Few-shot CoT - Clarification Helpfulness | 40.75 | Helpfulness (%) | 75 |
| CLAMBER - Zero-shot - Identification F1 | 50.99 | F1 (%) | 75 |
| Big-Bench Hard | 58.2 | Average (%) | 64.6 |
| OpenCompass Language - Content Summarization - English (CompassBench 2404) | 30 | Score (%) | 63.3 |
| OpenCompass Reasoning - Deductive - English (CompassBench 2404) | 13.9 | Score (%) | 63.3 |
| SAD - Self-Recognition (Situating Prompt) | 56.88 | Score (%, higher is better) | 60 |
| SAD - Count Tokens | 10.93 | Token-count accuracy (%, higher is better) | 59.1 |
| European LLM Leaderboard - FLORES200 | 11.71 | Average FLORES200 Score | 58.3 |
| European LLM Leaderboard - FLORES200 Target | 11.71 | FLORES200 Target Score | 58.3 |
Interactive version: theaggregate.ai/model?slug=llama-2-13b-chat · How It Works · Data refreshed daily, snapshot 2026-10-20.