Llama 3.1 405B Instruct
Zero-eval
#1ARC-C
#1API-Bank
#1Multilingual MGSM (CoT)
+7 more
by Meta
+
+
+
+
About
Llama 3.1 405B was developed as Meta's largest open-source language model, designed to provide frontier-level capabilities with 405 billion parameters. Built to demonstrate that open-source models can match proprietary systems in capability, it enables researchers and developers to experiment with and deploy a powerful foundation model without licensing restrictions.
+
+
+
+
Pricing Range
Input (per 1M)$0.89 -$9.50
Output (per 1M)$0.89 -$16.00
Providers8
+
+
+
+
Timeline
AnnouncedJul 23, 2024
ReleasedJul 23, 2024
+
+
+
+
Specifications
Training Tokens15.0T
+
+
+
+
License & Family
License
Llama 3.1 Community License
Performance Overview
Performance metrics and category breakdown
Overall Performance
18 benchmarks
Average Score
79.2%
Best Score
96.9%
High Performers (80%+)
11Performance Metrics
Max Context Window
256.0KAvg Throughput
48.3 tok/sAvg Latency
0ms+
+
+
+
All Benchmark Results for Llama 3.1 405B Instruct
Complete list of benchmark scores with detailed information
| ARC-C | text | 0.97 | 96.9% | Self-reported | |
| GSM8k | text | 0.97 | 96.8% | Self-reported | |
| API-Bank | text | 0.92 | 92.0% | Self-reported | |
| Multilingual MGSM (CoT) | text | 0.92 | 91.6% | Self-reported | |
| HumanEval | text | 0.89 | 89.0% | Self-reported | |
| MMLU (CoT) | text | 0.89 | 88.6% | Self-reported | |
| IFEval | text | 0.89 | 88.6% | Self-reported | |
| MBPP EvalPlus | text | 0.89 | 88.6% | Self-reported | |
| BFCL | text | 0.89 | 88.5% | Self-reported | |
| MMLU | text | 0.87 | 87.3% | Self-reported |
Showing 1 to 10 of 18 benchmarks
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+