Can NVIDIA DGX Spark 64GB run gpt-oss 20B?
Measured results for gpt-oss 20B (20.91B MoE, 3.6B active) on NVIDIA DGX Spark 64GB, plus what fits and the commands to run it.
Short answer
Yes. gpt-oss 20B Q8_0 fits fully on NVIDIA DGX Spark 64GB at 16k context, using 12.5 of 53.9 GB usable, and generates about 86.2 tokens/s. For Q8_0, our bandwidth-only estimate was ~72 tok/s against a measured median of ~86 (16% too pessimistic).
Measured generation speed by quantization
| Quant | Generation tok/s | Range | Prompt tok/s | Runs | Origin |
|---|---|---|---|---|---|
| Q4 class (median) | 86.2 | 83.4–117.8 | 3798 | 7 | public runs |
Generation = tg128 (128 generated tokens, short context). Public rows are medians of community llama.cpp results (sources); our own runs use llama-bench on a rented GPU.
Alternatives
Less memory that still fits gpt-oss 20B
Smaller memory usually means a cheaper card or Mac (we do not list prices).
- RTX 5080 16GB: Q8_0, 15 GB usable, ~239 tok/s measured
- RTX 5070 Ti 16GB: Q8_0, 15 GB usable, ~224 tok/s measured
- RTX 5060 Ti 16GB: Q8_0, 15 GB usable, ~125 tok/s measured
Faster for gpt-oss 20B
- RTX 4090 24GB: Q8_0, 23 GB usable, ~258 tok/s measured
- AMD RX 7900 XTX 24GB: Q8_0, 23 GB usable, ~221 tok/s estimated
- RTX 3090 24GB: Q8_0, 23 GB usable, ~204 tok/s measured
Other models on NVIDIA DGX Spark 64GB, similar size
- LFM2.5 24B-A2B (hardware page): Q8_0, fits fully, ~66 tok/s (scaled from measured)
- Mistral Small 3.2 24B (hardware page): Q4_K_M, fits fully, ~11 tok/s (scaled from measured)
- Gemma 4 26B-A4B (hardware page): Q8_0, fits fully, ~54 tok/s (scaled from measured)
- Qwen3.6 27B (hardware page): Q4_K_M, fits fully, ~13 tok/s (estimated)
Memory and context limits
Q8_0 weights are 12.1 GB; the 16k f16 KV cache adds 0.41 GB. Largest context that fits in 53.9 GB usable:
| KV cache type | Max context |
|---|---|
| f16 | 128k |
| q8_0 | 128k |
| q4_0 | 128k |
Native limit 128k. Full KV cache table for gpt-oss 20B.
Copy-paste commands
The advisor's pick for this hardware: Q8_0, 16k context.
llama-server -hf unsloth/gpt-oss-20b-GGUF:Q8_0 -c 16384 -fa on --jinja -ngl 99
OLLAMA_CONTEXT_LENGTH=16384 ollama serve ollama run hf.co/unsloth/gpt-oss-20b-GGUF:Q8_0
FAQ
How fast is gpt-oss 20B on NVIDIA DGX Spark 64GB?
About 86.2 tokens/s for generation (Q4 class, median of 7 public runs). Speed drops as the context fills; see the depth table on this hardware's other pages.
How much memory does gpt-oss 20B need on NVIDIA DGX Spark 64GB?
The advisor's pick, Q8_0, is 12.1 GB of weights plus 0.41 GB of KV cache at 16k context (12.5 GB), against 53.9 GB usable.
Is the gpt-oss 20B speed on NVIDIA DGX Spark 64GB measured or estimated?
Measured: the median of 7 public llama.cpp results for the Q4 quant class.
What can I use instead of NVIDIA DGX Spark 64GB for gpt-oss 20B?
With less memory, RTX 5080 16GB (~239 tok/s), RTX 5070 Ti 16GB (~224 tok/s), RTX 5060 Ti 16GB (~125 tok/s) also fit it fully. Faster: RTX 4090 24GB (~258 tok/s), AMD RX 7900 XTX 24GB (~221 tok/s), RTX 3090 24GB (~204 tok/s).
What is the maximum context for gpt-oss 20B on NVIDIA DGX Spark 64GB?
With an f16 KV cache about 128k tokens, 128k with q8_0 and 128k with q4_0, bounded by the model's own limit.
Related
All models on NVIDIA DGX Spark 64GB · gpt-oss 20B on all hardware · VRAM tier · all measured pages · leaderboard