2× DGX Spark
33tok/sPeak
1 request, best value reported by source, context 1m Source
Intelligence28with thinking
- Custom kernel required
- License unclear
Local AI · Recipe · 2× DGX Spark
Inkling-Small NVFP4 with vLLM on 2× DGX Spark
Inkling-Small NVFP4 with vLLM on 2× DGX Spark: 33 tok/s according to github.com (dataset as of Nov 3, 2026).
by tonyd2wild
Engine
- Engine
- vLLM eugr/spark-vllm @sha256:adc80460cea2…
- Quantization
- NVFP4 weights (ModelOpt) + BF16/auto KV
- Model family
- Inkling-Small
- Context
- 262,144
- Creator
- tonyd2wild
- GitHub stars
- 0
- Repo updated
- Jul 31, 2026
Measurements
Every sourced value of this recipe, each with its condition and source. Bars relative to the largest value in the group.
Decode · 1 request
Time to first token
Shorter is better.
What you need
- Hardware
- 2 × NVIDIA DGX Spark (GB10)
- Weights
- Main weightsthinkingmachines/Inkling-Small-NVFP4 Apache-2.0
- MTPthinkingmachines/Inkling-Small-NVFP4 Apache-2.0
- Engine
- vLLM eugr/spark-vllm @sha256:adc80460cea2…
- Context
- 262,144 tokens
Notes
What matters before you rebuild it.
- No license
Sources
Related recipes
2× DGX Spark
16tok/sPeak
1 request, mixed prompt set Source
Intelligence28with thinking
2× DGX Spark
no documented speed
Intelligence28with thinking
- Experimental
- License unclear
2× DGX Spark
89.3tok/sEveryday
1 request, prose prompt, with MTP (per recipe) Source
Intelligenceno independent value
- Custom kernel required
- License unclear
2× DGX Spark
74tok/sEveryday
1 request, prose prompt (long story), thinking off, with MTP Source
Intelligence40with thinking
- Custom kernel required
- License unclear
2× DGX Spark
69.6tok/sEveryday
1 request, prose prompt, with MTP (per recipe) Source
Intelligence40with thinking
- License unclear
Local AI in your company?
In a workshop we work out which models and which hardware fit your tasks, and build the first agent on your infrastructure.