Side-by-side comparison of DeepSeek V3 (DeepSeek) and Llama 3.1 405B (Meta) — benchmarks, pricing, and capabilities.
| DeepSeek V3 DeepSeek | Llama 3.1 405B Meta | |
|---|---|---|
| Category | LLMs | LLMs |
| Specifications | ||
| Context Window | 128K | 128K |
| Pricing (per 1M tokens) | ||
| Input Cost | $0.27 | Free |
| Output Cost | $1.10 | Free |
| Performance | ||
| Overall Score | 90.0 | 88.4 |
| ARC-Challenge | — | 89.7 |
| BigBench Hard | 85.5 | 82.3 |
| Chatbot Arena ELO | — | 1285.0 |
| DROP | 85.0 | 82.0 |
| GSM8K | 93.0 | 91.0 |
| HumanEval | 89.2 | 85.0 |
| MATH | 75.8 | 70.1 |
| MMLU | 87.1 | 84.2 |
| TruthfulQA | 68.0 | 55.0 |
| WinoGrande | — | 85.4 |
| Community | ||
| User Rating | ★ 4.5 | ★ 4.4 |
| Reviews | 620 | 420 |