Can DGX Spark (GH200 480GB) run Kimi-K2.7-Code?

NVIDIA · 480 GB LPDDR5X · 500 GB/s bandwidth · Moonshot · 1030.0B (active 32.0B) · MoE · model ctx up to 256K

CUDA

No — Kimi-K2.7-Code at Q4 needs ≈ 570.7 GB even at 4K context, beyond the DGX Spark (GH200 480GB)'s 480 GB LPDDR5X. It would only run with heavy CPU/disk offload. Decode at Q4/8K: ≈ 19.9 tok/s (estimated, batch 1). FP16 needs ≈ 2067.4 GB — does not fit on this card. Estimates come from memory-bandwidth math; rows tagged measured override estimates. Relative ranking is more reliable than absolute tok/s. Methodology.

Fit & speed by quant and context

Quant	Context	VRAM needed	Fits?	tok/s (decode)
Q4	4K	570.7 GB	No	20.1 ± est.
Q4	8K default	573.9 GB	No	19.9 ± est.
Q4	32K	593.1 GB	No	18.6 ± est.
FP16	4K	2064.2 GB	No	—
FP16	8K default	2067.4 GB	No	—
FP16	32K	2086.6 GB	No	—

Fits = weights + KV(ctx) + 1 GB overhead ≤ 95% of VRAM. Measured anchors are context-agnostic; the fit verdict is recomputed per context. A missing tok/s means the model is far beyond this card (offload-only territory).

VRAM breakdown at 8K context

Quant	Weights	KV cache	Overhead	Total needed	DGX Spark (GH200 480GB) VRAM
Q4	566.5 GB	6.4 GB	1 GB	573.9 GB	480 GB
FP16	2060.0 GB	6.4 GB	1 GB	2067.4 GB	480 GB

Other GPUs that run Kimi-K2.7-Code

No catalog GPU fully fits Kimi-K2.7-Code at Q4/8K — see the model page for offload estimates. model page

All GPUs for Kimi-K2.7-Code →

Other models for the DGX Spark (GH200 480GB)

• MiniCPM5-1B — ≈294.6 tok/s Q4 est.
• Qwen3.5-2B — ≈159.1 tok/s Q4 est.
• DeepSeek-R1-0528-Qwen3-8B-layer-mix-bpw-3.8-mlx — ≈138.3 tok/s Q4 est.
• Qwen3.6-35B-A3B — ≈106.1 tok/s Q4 est.
• Qwen3.5-35B-A3B — ≈106.1 tok/s Q4 est.
• GLM-4.7-Flash — ≈106.1 tok/s Q4 est.
• Qwen3-30B-A3B-Instruct-2507 — ≈106.1 tok/s Q4 est.
• Nemotron-3-Nano-Omni-30B-A3B-Reasoning-BF16 — ≈106.1 tok/s Q4 est.

All models on the DGX Spark (GH200 480GB) →