Open-weight LLM comparison
Llama 3.1 70B vs Llama 3.3 70B
Side-by-side
Straight from each model's release page and Hugging Face card. Memory shown is for the weights alone: add roughly 15% for real serving. Bold column marks the more parameters and the smaller memory footprint.
| Field | Llama 3.1 70B | Llama 3.3 70B |
|---|---|---|
| Total parameters | 70 B | 70 B |
| Architecture | dense | dense |
| Vendor | Meta | Meta |
| License | Llama 3.1 Community License | Llama 3.3 Community License |
| Released | 2024-07-23 | 2024-12-06 |
| Weights @ FP16 | 140 GB | 140 GB |
| Weights @ INT8 | 70 GB | 70 GB |
| Weights @ INT4 | 35 GB | 35 GB |
Which is smarter (published benchmarks)
Vendor-reported quality scores on the standard leaderboards. Bold column marks the higher score on the same test.
| Benchmark | Llama 3.1 70B | Llama 3.3 70B |
|---|---|---|
| MMLU (5-shot) | 86.0 | 86.0 |
| HumanEval | 80.5 | 88.4 |
| MATH (0-shot) | 68.0 | 77.0 |
| GPQA | 46.7 | 50.5 |
Sources: Llama 3.1 70B model card · Llama 3.3 70B model card. Benchmark methodology and prompt template can shift these numbers by several points, so treat these as relative rankings, not absolute scores.
Common questions
Llama 3.1 70B vs Llama 3.3 70B: which is newer?
Llama 3.3 70B released 2024-12-06; Llama 3.1 70B released 2024-07-23.
Llama 3.1 70B vs Llama 3.3 70B: which license is more permissive?
Llama 3.1 70B: Llama 3.1 Community License. Llama 3.3 70B: Llama 3.3 Community License. Apache 2.0 and MIT allow unrestricted commercial use; Llama Community License allows commercial use but restricts training larger models on outputs; CC-BY-NC and vendor-specific licenses (Qwen 72B, Gemma) have narrower terms. Check the model card for the exact clauses.
See also: every LLM comparison · Llama 3.1 70B full page · Llama 3.3 70B full page.