← CanIRunLocal · All hardware · All models

Can I run DeepSeek R1 Distill Qwen 1.5B locally?

DeepSeek · 1.78B parameters · reasoning.

Ollama weight files: Q4_K_M 1.1 GB (deepseek-r1:1.5b); Q8_0 1.9 GB (deepseek-r1:1.5b-qwen-distill-q8_0).

Yes on 62 machines in this list. The smallest is M1 8GB: Q8_0, about 2.6 GB, roughly 29 tokens per second, via Ollama. ollama run deepseek-r1:1.5b-qwen-distill-q8_0

• DGX Spark 128GB: Q8_0, ~2.6 GB, fits in fast memory, ~108 tok/s — ollama run deepseek-r1:1.5b-qwen-distill-q8_0
• Mac Studio M5 Max 128GB: Q8_0, ~2.6 GB, fits in fast memory, ~259 tok/s — ollama run deepseek-r1:1.5b-qwen-distill-q8_0
• Mac Studio M5 Ultra 256GB: Q8_0, ~2.6 GB, fits in fast memory, ~505 tok/s — ollama run deepseek-r1:1.5b-qwen-distill-q8_0

Setup on DGX Spark 128GB (NVIDIA, ASUS, Dell, Lenovo, HP, MSI, Acer, Gigabyte)

• Ollama: ollama run deepseek-r1:1.5b-qwen-distill-q8_0
• LM Studio: lms get unsloth/DeepSeek-R1-Distill-Qwen-1.5B-GGUF@Q8_0
• llama.cpp: llama-server -hf unsloth/DeepSeek-R1-Distill-Qwen-1.5B-GGUF:Q8_0 -c 8192
• vLLM: vllm serve deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B --max-model-len 8192 --gpu-memory-utilization 0.9

Hardware that fits

Estimates, not benchmarks. Model sizes are ollama.com download sizes; memory needed adds headroom for ~8k tokens of context; speeds are rough figures from memory bandwidth. Data updated 2026-10-16.

A project by OnDevice Labs · llms.txt