Meta AI·released 2023-07-181 source
Llama 2-13B benchmark scores: 13 benchmarks tracked. 0% of its scores are independently reproduced — the rest are self-reported or unverified.
Leads on 0 of 13 benchmarks·best result 79.6 on TriviaQA (unverified)·0 of 13 independently reproduced
| Benchmark | Score | Evidence status | Measured |
|---|---|---|---|
| TriviaQA | 79.6accuracy (%) | unverified· optimizedT1 | 2023-07-18Gemma: Open Models Based on Gemini Research and Technology |
| LAMBADA | 76.5accuracy (%) | self-reported· optimizedT1 | 2023-07-18Qwen Technical Report |
| HellaSwag | 74.27accuracy (%) | unverified· optimizedT1 | 2023-07-18Gemma: Open Models Based on Gemini Research and Technology |
| PIQA | 61.6accuracy (%) | unverified· optimizedT1 | 2023-07-18Gemma: Open Models Based on Gemini Research and Technology |
| ARC AI2 | 47.07accuracy (%) | unverified· optimizedT1 | 2023-07-18The Falcon Series of Open Language Models |
| Winogrande | 45.6accuracy (%) | unverified· optimizedT1 | 2023-07-18Llama 2: Open Foundation and Fine-Tuned Chat Models |
| BBH | 44.27accuracy (%) | unverified· optimizedT1 | 2023-07-18Gemma: Open Models Based on Gemini Research and Technology |
| OpenBookQA | 42.67accuracy (%) | unverified· optimizedT1 | 2023-07-18Llama 2: Open Foundation and Fine-Tuned Chat Models |
| ScienceQA | 41.04accuracy (%) | unverified· optimizedT1 | 2023-07-18ScienceQA leaderboard |
| MMLU | 40.8accuracy (%) | unverified· optimizedT1 | 2023-07-18Stanford CRFM Leaderboard |
| GSM8K | 36.9accuracy (%) | unverified· optimizedT1 | 2023-07-18Stanford HELM |
| DTBench | 3.73accuracy (%) | unverifiedT1 | 2023-07-18https://conceptualreasoning.ai/dtbench |
| Chess Puzzles | 0accuracy (%) | unverifiedT2 | 2023-07-18 |
Claims drawn from cited facts, not live model generation.
This model was developed by Meta AI and released in 2023, placing it in the mid-size tier with 13 billion parameters. That mid-size scale is a deliberate trade-off: it offers substantially more headroom than compact models while remaining far cheaper and more efficient to serve than frontier-scale systems.
4 cited facts
This model is scored on 13 tracked benchmarks, but it currently holds no top score on any of them. Its benchmark record is self-reported, meaning the numbers are vendor-claimed and have not yet been independently confirmed, so they should be read as claims rather than verified results.
3 cited facts
On average the model trails the benchmark leader by 39.74 points, a measured score gap under the disclosed harness rather than a capability claim; that wide gap places it well behind the leaders. It also holds the top score on 0 benchmarks, which means no genuine category leadership yet.
2 cited facts
2 facts cross-checked across data sources: 2 single-source.
Source facts, citations, and refresh stamp for this record.
Sources: curated_param_counts, epoch_benchmarks
Llama 2-13B is an AI model developed by Meta AI, released 2023-07-18. Its benchmark record below tags every score as independently reproduced, self-reported, or unverified.
Llama 2-13B has recorded scores on 13 benchmarks, each shown with its evidence status.
Llama 2-13B has recorded scores on 13 benchmarks — PIQA, OpenBookQA, Winogrande, TriviaQA, DTBench, MMLU, and 7 more. The full table above shows each score with its evidence status.
0 of 13 recorded scores (0%) are independently reproduced rather than self-reported by the lab.
Llama 2-13B has 13 tracked claims: 1 self-reported, 12 unverified.