Skip to content

Kimi K2 Instruct on Dual DGX Spark: VRAM, speed and fit

Kimi K2 Instruct at 1.6-bit runs on the DGX Spark 128 GB with 128K of context at about 21 tokens a second.

Text generation · Chat · Ollama / LM Studio

Runs comfortably · on DGX Spark 128 GB · 1.6-bit (IQ1_S) · Ollama / LM Studio · Chat

GPU memory
222 GB
of 256 GB
Context
128K
auto
Speed
21 tok/s
10 when full
First reply
5.3 s
2K prompt

Where the memory goes

Model
197 GB
Context
8.6 GB
Overhead
1.3 GB
OS
15.4 GB
Free
33.6 GB
Memory
222 GB of 256 GB
Context
128K tokens · auto
Speed
21 tok/s · 10 when full
First reply
5.3 s for a 2K prompt
Running cost
$3.15/mo at home · $574/mo on a cloud 8× A40

Validation: Validated with conditions

What Kimi K2 Instruct on Dual DGX Spark was checked against
CheckTargetResultStatus
Fits in GPU memory241 GB usable207 GB neededPass
Headroom5% free33.6 GB free (14%)Pass
Context for chat16K or more128KPass
Writing speed15 tok/s or more21 tok/sPass
First replyunder 3 s, 2K prompt5.3 sCheck

Targets: a reply you can read along with, a first reply the use case tolerates, and 5% of memory left free. Calculated from published specs, not benchmarked.

The setup

Model
Kimi K2 Instruct
Parameters
1T, 41.5B active
Precision
1.6-bit (IQ1_S)
Context
128K, auto
Chats at once
1
Use
Chat
Runs in
Ollama / LM Studio
Hardware
DGX Spark 128 GB
GPU can use
241 GB of 256 GB

Running cost

At home
$3.15/mo
144 W at the wall
Renting instead
$574/mo
8× A40 at $4.72/h

4 h a day, power at $0.18 per kWh, cloud at RunPod on-demand prices as of Nov 21, 2026.

How to run it

llama.cpp
llama-server -hf unsloth/Kimi-K2-Instruct-GGUF
vLLM
vllm serve moonshotai/Kimi-K2-Instruct
MLX (Mac)
mlx_lm.chat --model mlx-community/Kimi-K2-Instruct-4bit

Kimi K2 Instruct on similar machines

All machines: Kimi K2 Instruct VRAM requirements. Everything that runs here: what can the Dual DGX Spark run?

Other models that run on Dual DGX Spark

Questions

Can Kimi K2 Instruct run on Dual DGX Spark?

Kimi K2 Instruct at 1.6-bit runs on the DGX Spark 128 GB with 128K of context at about 21 tokens a second.

Report VR-1D8LVZC, calculated Nov 22, 2026 from published specs, not benchmarked.