Meta AI·released 2024-04-181 source
Llama 3-8B benchmark scores: 11 benchmarks tracked. 18% of its scores are independently reproduced — the rest are self-reported or unverified.
Leads on 0 of 11 benchmarks·best result 77.07 on ARC AI2 (self-reported)·2 of 11 independently reproduced
| Benchmark | Score | Evidence status | Measured |
|---|---|---|---|
| ARC AI2 | 77.07accuracy (%) | self-reported· optimizedT1 | 2024-04-18Phi-3 Technical Report: A Highly Capable Language Model Locally on Your Phone |
| OpenBookQA | 76.8accuracy (%) | self-reported· optimizedT1 | 2024-04-18Phi-3 Technical Report: A Highly Capable Language Model Locally on Your Phone |
| TriviaQA | 67.7accuracy (%) | self-reported· optimizedT1 | 2024-04-18Phi-3 Technical Report: A Highly Capable Language Model Locally on Your Phone |
| MMLU | 58.4accuracy (%) | unverified· optimizedT1 | 2024-04-18Stanford CRFM Leaderboard |
| Winogrande | 51.4accuracy (%) | unverified· optimizedT1 | 2024-04-18The Llama 3 Herd of Models |
| ANLI | 35.95accuracy (%) | self-reported· optimizedT1 | 2024-04-18Phi-3 Technical Report: A Highly Capable Language Model Locally on Your Phone |
| DTBench | 6.55accuracy (%) | unverifiedT1 | 2024-04-18https://conceptualreasoning.ai/dtbench |
| MATH level 5 | 6.13accuracy (%) | reproduced· optimizedT1 | 2024-04-18Epoch AI |
| OTIS Mock AIME 2024-2025 | 1.85accuracy (%) | unverifiedT2 | 2024-04-18 |
| GPQA diamond | 1.43accuracy (%) | reproduced· optimizedT1 | 2024-04-18Epoch AI |
| Chess Puzzles | 0accuracy (%) | unverifiedT2 | 2024-04-18 |
Claims drawn from cited facts, not live model generation.
This model was developed by Meta AI and released in 2024 at a compact 8 billion parameters. That modest scale cuts inference cost and keeps the model lightweight to serve, at the price of less headroom than larger models can offer.
4 cited facts
This model is scored on 11 tracked benchmarks and currently holds no top score. Its results are independently reproduced, meaning outside evaluation backs the scores up.
3 cited facts
On average this model trails the leading score by 49.28 points, a gap measured under a disclosed harness that reads as well behind the front of the pack rather than merely competitive. Consistent with that wide margin, it holds the top score on 0 benchmarks, meaning it has no category leadership on any of its evaluated benchmarks yet.
3 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 3-8B is an AI model developed by Meta AI, released 2024-04-18. Its benchmark record below tags every score as independently reproduced, self-reported, or unverified.
Llama 3-8B has recorded scores on 11 benchmarks, each shown with its evidence status.
Llama 3-8B has recorded scores on 11 benchmarks — OpenBookQA, Winogrande, GPQA diamond, MATH level 5, OTIS Mock AIME 2024-2025, TriviaQA, and 5 more. The full table above shows each score with its evidence status.
2 of 11 recorded scores (18%) are independently reproduced rather than self-reported by the lab.
Llama 3-8B has 11 tracked claims: 2 independently reproduced, 4 self-reported, 5 unverified.