Meta AI·released 2024-09-241 source
Llama 3.2 90B benchmark scores: 6 benchmarks tracked. 50% of its scores are independently reproduced — the rest are self-reported or unverified.
Leads on 0 of 6 benchmarks·best result 73.73 on MMLU (unverified)·3 of 6 independently reproduced
| Benchmark | Score | Evidence status | Measured |
|---|---|---|---|
| MMLU | 73.73accuracy (%) | unverified· optimizedT1 | 2024-09-24Stanford CRFM Leaderboard |
| GeoBench | 52accuracy (%) | unverifiedT1 | 2024-09-24GeoBench leaderboard |
| MATH level 5 | 39.44accuracy (%) | reproduced· optimizedT1 | 2024-09-24Epoch AI |
| Balrog | 27.3accuracy (%) | unverifiedT1 | 2024-09-24https://balrogai.com/ |
| GPQA diamond | 21.38accuracy (%) | reproduced· optimizedT1 | 2024-09-24Epoch AI |
| OTIS Mock AIME 2024-2025 | 2.54accuracy (%) | reproduced· optimizedT1 | 2024-09-24Epoch AI |
Claims drawn from cited facts, not live model generation.
Developed by Meta AI and released in 2024, this model is a frontier-scale system with 90.0 billion parameters; that scale affords considerable headroom for complex tasks, yet also brings higher compute and deployment costs than a smaller model would.
3 cited facts
This model is scored on 6 tracked benchmarks and currently holds no top score on any of them. Because these results have been independently reproduced, the record can be trusted as verified.
3 cited facts
With an average SOTA gap of 50.84 points under the disclosed harness, this model trails the leader by a wide margin and is well behind the front-of-pack; it holds the top score on 0 benchmarks, meaning no 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 3.2 90B is an AI model developed by Meta AI, released 2024-09-24. Its benchmark record below tags every score as independently reproduced, self-reported, or unverified.
Llama 3.2 90B has recorded scores on 6 benchmarks, each shown with its evidence status.
Llama 3.2 90B has recorded scores on 6 benchmarks — GPQA diamond, MATH level 5, OTIS Mock AIME 2024-2025, MMLU, Balrog, GeoBench. The full table above shows each score with its evidence status.
3 of 6 recorded scores (50%) are independently reproduced rather than self-reported by the lab.
Llama 3.2 90B has 6 tracked claims: 3 independently reproduced, 3 unverified.