Llama 3.1 70B Instruct
Open Sourcedi Meta · Rilascio 2024-07-23
37.8
punteggio medio
$0.40/1M
Prezzo Input
$0.40/1M
Prezzo Output
131K tokens (~66 books)
Finestra di Contesto
text
Tipo
Tested on 16 benchmarks with 37.8% average. Top scores: Chatbot Arena Elo — Overall (1292.8%), IFEval (86.7%), MMLU (73.5%).
Punteggi Benchmark
| Benchmark | Categoria | Punteggio | Bar |
|---|---|---|---|
| Chatbot Arena Elo — Overall | arena | 1292.8 | |
| IFEval | language | 86.7 | |
| MMLU | knowledge | 73.5 | |
| CMMLU | knowledge | 64.4 | |
| Aider — Code Editing | coding | 58.6 | |
| BBH (HuggingFace) | general | 55.9 | |
| MMLU-PRO | knowledge | 47.9 | |
| MATH Level 5 | math | 38.1 | |
| MATH level 5 | math | 36.7 | |
| Balrog | knowledge | 27.9 | |
| GPQA diamond | knowledge | 25.6 | |
| MUSR | reasoning | 17.7 | |
| GPQA | knowledge | 14.2 | |
| WeirdML | coding | 9.0 | |
| The Agent Company | agentic | 6.9 | |
| OTIS Mock AIME 2024-2025 | math | 3.5 |