Only with CPU offload

DeepSeek-R1-Distill-Qwen 7B at Q4_K_M needs 5.3 GB but only 5.3 GB is addressable, so about 1% of the layers would stream from system RAM at roughly 60 GB/s. Expect around 46 tokens per second — usable for batch work, painful for chat.

99% on GPU 8K context Q4_K_M · 4.3 GB MIT Released Jan 2025

Reasoning traces on a 7B budget. Expect long outputs — budget context accordingly.

Hardware requirements What hardware do I need to run DeepSeek-R1-Distill-Qwen 7B? →

At Q4_K_M with an 8K context: a 8 GB GPU (RTX 4060 Laptop) or a 16 GB Mac (M4). Minimum, recommended, high-quality and long-context tiers, with all 114 devices checked at every quantisation.

The VRAM budget

weights 4.3 GB
Weights 4.3 GB KV cache @ 8K 0.44 GB Runtime overhead 0.6 GB Over budget 0.0 GB past 5.3 GB

Quantisation ladder

QuantWeightsTotal @ 8KMax contextTok/sQualityFit
F16 14.2 GB 15.2 GB — ~3.4 Reference 9.9 GB over
Q8_0 7.5 GB 8.6 GB — ~9.0 −0.1% ppl 3.3 GB over
Q6_K 5.8 GB 6.9 GB — ~16 −0.4% ppl 1.6 GB over
Q5_K_M 5.0 GB 6.1 GB — ~24 −0.8% ppl 0.8 GB over
Q4_K_M 4.3 GB 5.3 GB 7K ~46 −1.9% ppl 0.0 GB over
Q3_K_M 3.5 GB 4.5 GB 22K 59 −5.4% ppl Long context
Q2_K 3.0 GB 4.0 GB 31K 68 −15% ppl Long context

Quality is the published perplexity delta against f16 weights. Max context assumes an f16 KV cache; q8_0 roughly doubles it.

How to run it

terminal
$ llama-server \
    -hf deepseek-ai/DeepSeek-R1-Distill-Qwen-7B:Q4_K_M \
    -c 8192 -ngl 27

The engine underneath most of the others. Every knob is exposed. More on llama.cpp.

01Download is 4.3 GB. Keep it on an SSD — a first load off a spinning disk takes minutes.
02Close anything else holding VRAM. A browser with hardware acceleration can sit on 1–2 GB.
03If it falls back to CPU silently, drop the context first, then step down a quantisation.
See all models for this rig Compare with another model

Related models on a GeForce GTX 1660 Super

ModelTotalTok/s est.
DeepSeek-R1-Distill-Qwen 14B
14.8B · same family
Q4_K_M · 10.4 GB ~6.4 CPU offload
DeepSeek-R1-Distill-Qwen 32B
32.8B · same family
Q4_K_M · 21.0 GB ~2.2 CPU offload
DeepSeek-R1-Distill-Llama 70B
70.6B · same family
Q4_K_M · 42.8 GB ~1.0 CPU offload
DeepSeek-R1 671B
671B MoE · same family
Q4_K_M · 378.4 GB — Won't fit
Qwen2.5-Coder 7B
7.62B · similar size
Q4_K_M · 5.3 GB ~46 CPU offload
Ling 3.0 Tiny 7.9B-A1.3B
7.9B MoE · similar size
Q4_K_M · 5.1 GB 107 Tight fit
Olmo 3 7B Instruct
7.3B · similar size
Q4_K_M · 7.2 GB ~16 CPU offload

Every model against the GeForce GTX 1660 Super →

DeepSeek-R1-Distill-Qwen 7B on other hardware

DeviceTotalTok/s est.
GeForce RTX 4060 Laptop
7.0 GB usable · 256 GB/s
5.3 GB 36 Runs great
GeForce RTX 3060 12 GB
10.6 GB usable · 360 GB/s
5.3 GB 51 Runs great
M4 · 16 GB
10.7 GB usable · 120 GB/s
5.3 GB 16 Runs great
GeForce RTX 4090
22.4 GB usable · 1008 GB/s
5.3 GB 142 Runs great
Radeon RX 7900 XTX
22.4 GB usable · 960 GB/s
5.3 GB 136 Runs great
M3 Max · 64 GB
48.0 GB usable · 400 GB/s
5.3 GB 52 Runs great

All 114 devices, by tier →