DeepSeek R1 7B
by DeepSeek · deepseek-r1 family
7B
parameters
text-generation code-generation reasoning math
DeepSeek R1 7B is a distilled reasoning model from DeepSeek's R1 series, based on the Qwen 2.5 architecture. It inherits chain-of-thought reasoning capabilities from the larger DeepSeek R1 model through knowledge distillation, delivering strong reasoning performance at a compact size. This model excels at mathematical problem solving, logical reasoning, and coding tasks. It shows its reasoning process step by step, making it transparent and useful for educational and analytical purposes. It runs efficiently on consumer hardware.
Quick Start with Ollama
ollama run 7b-q8_0 | Creator | DeepSeek |
| Parameters | 7B |
| Architecture | transformer-decoder |
| Context Length | 128K tokens |
| License | MIT |
| Released | Jan 20, 2025 |
| Ollama | deepseek-r1:7b |
Quantization Options
| Format | File Size | VRAM Required | Quality | Ollama Tag |
|---|---|---|---|---|
| Q4_K_M | 3.5 GB | 5.7 GB |
★
★
★
★
★
| 7b-q4_K_M |
| Q8_0 recommended | 6.3 GB | 9 GB |
★
★
★
★
★
| 7b-q8_0 |
| F16 | 13.3 GB | 16 GB |
★
★
★
★
★
| 7b-fp16 |
Compatible Hardware for Q8_0
Showing compatibility for the recommended quantization (Q8_0, 9 GB VRAM).
Compatible Hardware
Benchmark Scores
68.5
mmlu