DeepSeek: R1 Distill Llama 70B
DeepSeek R1 Distill Llama 70B shows strong math performance but modest scores across reasoning and coding benchmarks overall. It's a cost-effective distilled reasoning model, though limited Western API availability and compliance considerations are relevant for UK businesses.
Assessment date: March 12, 2026
Our methodology takes into account a range of factors including pricing, functionality, capabilities, benchmark performance, and real-world applicability. Rankings are reviewed and updated regularly as new models are released. Issues with our rankings? Contact us
DeepSeek R1 Distill Llama 70B is a distilled large language model based on Llama-3.3-70B-Instruct, using outputs from DeepSeek R1. The model combines advanced distillation techniques to achieve high performance across multiple benchmarks, including: - AIME 2024 pass@1: 70.0
- MATH-500 pass@1: 94.5
- CodeForces Rating: 1633 The model leverages fine-tuning from DeepSeek R1's outputs, enabling competitive performance comparable to larger frontier models.
Architecture
| Modality | Text → Text |
| Tokenizer | Llama3 |
| Instruct Type | deepseek-r1 |
| Parameters | 70B |
Performance Indices
Source: Artificial Analysis
Benchmark Scores
Evaluations
Benchmark data from Artificial Analysis and Hugging Face
Model Information
Pricing
| Token Type | Cost per 1M tokens | Cost per 1K tokens |
|---|---|---|
| Input | $0.70 | $0.000700 |
| Output | $0.80 | $0.000800 |
Live Performance
Live endpoint metrics — refreshed every 30 minutes.
