1× Mac
53.3tok/sEveryday
1 request, prose prompt, with MTP (nativ) (per recipe) Source
Intelligenceno independent value
- License unclear
Local AI · Recipe · 1× DGX Spark
Qwen3.8-27B NVFP4 with Atlas on 1× DGX Spark
Qwen3.8-27B NVFP4 with Atlas on 1× DGX Spark: 19 tok/s according to github.com (dataset as of Sep 29, 2026).
by Weschera
Intelligence (original model) · Artificial Analysisno independent value
Engine
- Engine
- Atlas
- Quantization
- NVFP4 (unsloth); MTP-Head bf16 (--mtp-quantization bf16); fp8 KV (Engine-Default)
- Model family
- Qwen3.8-27B
- Context
- 32,768
- Parameters
- 27B (dicht)
- Creator
- Weschera
- GitHub stars
- 0
- Repo updated
- Aug 15, 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
What you need
- Hardware
- 1 × NVIDIA DGX Spark (GB10)
- Weights
- Main weightsunsloth/Qwen3.8-27B-NVFP4 Apache-2.0
- Engine
- Atlas
- Context
- 32,768 tokens
Notes
What matters before you rebuild it.
- Experimental
- Custom kernel required
Sources
Related recipes
1× DGX Spark
45tok/sEveryday
1 request, prose prompt, with MTP Source
Intelligenceno independent value
4× DGX Spark
35.1tok/sEveryday
1 request, realistic prompt, with Qwen MTP3 probabilistic mit standard rejection (per recipe) Source
Intelligence34with thinkingno thinking 20
- Custom kernel required
1× RTX 3090
27.8tok/sMixed
1 request, mixed prompt set, context 150000, with DFlash2 (per recipe) Source
Intelligence34with thinkingno thinking 20
- Custom kernel required
1× DGX Spark
27.8tok/sMixed
1 request, mixed prompt set, with MTP Source
Intelligence34with thinkingno thinking 20
- License unclear
1× DGX Spark
25.2tok/sMixed
1 request, mixed prompt set, with DFlash2 (per recipe) Source
Intelligenceno independent value
- Custom kernel required
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.