Skip to content

What can the Dual DGX Spark run?

What fits on the Dual DGX Spark
Text models: 81Visual models: 0Strongest first
Memory
256 GB unified
Bandwidth
273 GB/s each
Maker
NVIDIA
Also called
2× DGX Spark

Open the calculator on this machineDGX Spark (GB10) in our inventory

Text models that run on it

Text models that run on the Dual DGX Spark, best first
ModelBest fitContextSpeedStrongest at
MiMo V2.6 Pro1.6-bit608K20 tok/sChat, Coding, Agents, Docs
DeepSeek V4.1 Flash1.6-bit1M48 tok/sChat, Coding, Agents, Docs
GLM-5.31.6-bit864K21 tok/sChat, Coding, Agents, Docs
GLM-5.3 Flash5-bit1M14 tok/sChat, Coding, Agents, Docs
Qwen3.8 Flash-Next8-bit256K27 tok/sChat, Coding, Agents
Muse Spark Open 816B A42B1.6-bit1M20 tok/sChat, Docs
MiMo V2.6 FlashFP4 + FP81M17 tok/sCoding, Agents
MiniMax M33-bit256K16 tok/sDocs
DeepSeek V4 FlashFP4 + FP81M9.9 tok/sDocs
Kimi K2.61.6-bit256K26 tok/sChat, Coding, Agents, Docs
Hy35-bit96K12 tok/sChat, Coding, Agents, Docs
Nemotron 3 Ultra 550B A55B2-bit256K8.6 tok/sChat, Coding, Docs
Qwen3.5 397B A17B3-bit256K21 tok/sAgents, Docs
MiniMax M2.7FP852K18 tok/sCoding, Agents
DeepSeek V3.21.6-bit160K23 tok/sChat, Coding, Agents, Docs
Mistral Medium 3.5 128BFP8256K1.4 tok/sChat, Coding, Agents, Docs
Qwen3.5 122B A10B8-bit256K17 tok/sChat, Coding, Agents, Docs
Nemotron 3 Super 120B A12B8-bit256K14 tok/sChat, Coding, Agents, Docs
Nemotron 3.5 Bolt 74B A9B8-bit256K18 tok/sChat, Coding, Agents, Docs
Mistral Small 4 119BFP8256K27 tok/sChat, Docs
gpt-oss 120BMXFP4128K47 tok/sChat, Coding
Qwen3 Coder Next 80B8-bit256K50 tok/sCoding, Agents
Llama 3.3 70B8-bit128K2.6 tok/sChat, Coding, Agents, Docs
Inkling1.6-bit4K21 tok/sText and images
DeepSeek R11.6-bit160K21 tok/sNot rated
DeepSeek V31.6-bit160K21 tok/sNot rated
GLM 51.6-bit198K20 tok/sNot rated
GLM 4.5 Air8-bit128K12 tok/sNot rated
Inkling Small6-bit4K19 tok/sText and images
MiniMax M2.5FP852K15 tok/sNot rated
GLM 5.11.6-bit198K20 tok/sNot rated
Laguna M.16-bit200K9.5 tok/sNot rated
Laguna S 2.18-bit1M20 tok/sNot rated
Qwen3 VL 235B A22B Instruct6-bit256K10.0 tok/sText and images
GLM 5.21.6-bit864K20 tok/sNot rated
Qwen3 235B A22B6-bit40K10.0 tok/sNot rated
Ornith 1.5 122B8-bit256K13 tok/sNot rated
Kimi K2 Instruct1.6-bit128K21 tok/sNot rated
Ornith 1.5 397B3-bit256K15 tok/sNot rated
Ornith 1.0 397B4-bit96K16 tok/sNot rated
Step 3.5 Flash8-bit72K18 tok/sNot rated
DeepSeek V3 03241.6-bit160K21 tok/sNot rated
Qwen2.5 72B Instruct8-bit32K2.5 tok/sNot rated
Qwen3 Next 80B A3B Instruct8-bit256K50 tok/sNot rated
DeepSeek R1 05281.6-bit160K21 tok/sNot rated
GLM 4.54-bit64K8.4 tok/sNot rated
Llama 4 Scout 17B 16E Instruct8-bit608K9.7 tok/sText and images
MiniMax M2FP852K15 tok/sNot rated
MiMo V2.55-bit960K15 tok/sNot rated
Qwen2.5 VL 72B Instruct8-bit125K2.5 tok/sText and images
Solar Open 100B8-bit128K14 tok/sNot rated
GLM 5.3 Flash BF165-bit1M14 tok/sText and images
Qwen3 VL 235B A22B Thinking6-bit256K10.0 tok/sText and images
Qwen3 235B A22B Instruct 25076-bit256K10.0 tok/sNot rated
NVIDIA Nemotron 3.5 Super VL 120B A12B BF168-bit1M14 tok/sText and images
Hy3 Preview5-bit88K12 tok/sNot rated
Qwen3 Coder 480B A35B Instruct3-bit36K11 tok/sCoding
GLM 4.74-bit64K8.4 tok/sNot rated
Qwen3 Next 80B A3B Thinking8-bit256K50 tok/sNot rated
GLM 5.3 BF161.6-bit864K20 tok/sNot rated

Best fit is the highest quality, up to 8-bit, that leaves the use case’s context. Ranked by our use-case scores, then by Hugging Face downloads. Speeds are rough, plus or minus 25%. The top 60 of 81 are shown.

Questions

What is the best LLM for the Dual DGX Spark?

Of the open models we track, the strongest that run well on the Dual DGX Spark are MiMo V2.6 Pro, DeepSeek V4.1 Flash, GLM-5.3. 81 text models fit it in all.

Calculated from published specs, not benchmarked. All machines