Meta
Llama 3.3 70B benchmark results
9 results on 8 benchmarks from 5 sources.
Also listed as: llama-3.3-70b-instruct · Llama-3.3-70B-Instruct · Llama 3.3 Instruct Turbo (70B) · Llama3.3 70b instruct · Llama-3.3-70B-Instruct (FC) · meta-llama/Llama-3.3-70B-Instruct
All benchmark results
| Benchmarks | Score | Rank | Setting | Source type | Date |
|---|---|---|---|---|---|
| LMArena Text | 1318 | 262 | Overall, style control on; 54504 votes; 95% CI 1315–1322“llama-3.3-70b-instruct” | Official leaderboard | Oct 8, 2026 |
| Open LLM Leaderboard v2 | 44.9% | 40 | Average of 6 normalised benchmarks; bfloat16; 70.554B params“meta-llama/Llama-3.3-70B-Instruct” | Official leaderboard | Mar 13, 2025 |
| GPQA Diamond | 47.4% | 217 | Epoch AI run, accuracy | Independent run | Jan 27, 2025 |
| MMLU | 86.3% | 4 | Exact match; 5-shot | Paper / tech report via Epoch AI | Dec 6, 2024 |
| MMLU | 79.1% | 23 | Exact match; 5-shot | Official leaderboard via Epoch AI | Dec 6, 2024 |
| MMLU-Pro | 65.9% | 115 | Overall accuracy; data source: TIGER-Lab; 70B“Llama-3.3-70B-Instruct” | Official leaderboard | Mar 11, 2026 |
| AIME (OTIS Mock AIME 2024–2025) | 5.1% | 233 | Epoch AI run, OTIS Mock AIME 2024–2025 | Independent run | Feb 25, 2025 |
| MATH Level 5 | 41.6% | 69 | Epoch AI run, accuracy | Independent run | Jan 27, 2025 |
| BFCL v4 (function calling) | 31.9% | 62 | Overall accuracy; native function calling (FC)“Llama-3.3-70B-Instruct (FC)” | Official leaderboard | Oct 17, 2026 |
Related on this site
FAQ
Where do these Llama 3.3 70B scores come from?
From Epoch AI Benchmarking Hub, Berkeley Function Calling Leaderboard (BFCL), LMArena leaderboard dataset, MMLU-Pro leaderboard (TIGER-Lab), and Hugging Face Open LLM Leaderboard v2 (archived). Each row links to the original leaderboard, dataset or report, and vendor-reported numbers are labelled.