4× DGX Spark
87.7tok/sEveryday
1 request, prose prompt, with DSpark (per recipe) Source
Intelligence40with thinkingno thinking 25
Local AI · Recipe · 4× RTX PRO 6000
DeepSeek-V4.1-Flash 284B MXFP4 with SGLang on 4× RTX PRO 6000
DeepSeek-V4.1-Flash 284B MXFP4 with SGLang on 4× RTX PRO 6000: 73.1 tok/s according to github.com (dataset as of Nov 3, 2026).
by ambientlight
Intelligence (original model) · Artificial Analysis40with thinkingno thinking 25
Engine
- Engine
- SGLang
- Quantization
- nativ: MXFP4-Experten + FP8 (offizielle Gewichte), FP8-KV
- Model family
- DeepSeek-V4.1-Flash
- Context
- 524,288
- Parameters
- 284B MoE
- Creator
- ambientlight
- GitHub stars
- 5
- Repo updated
- Nov 1, 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
Prefill by context
What you need
- Hardware
- 4 × NVIDIA RTX PRO 6000 Blackwell
- Weights
- Main weightsdeepseek-ai/DeepSeek-V4.1-Flash MIT
- Engine
- SGLang
- Context
- 524,288 tokens
Sources
- github.com/ambientlight/rtx-pro-6000-bench/blob/main/bench/deepseek-v4.1-flash_TP4_sglang/BENCHMARK-LONG-TPS-2026-09-15.md
- github.com/ambientlight/rtx-pro-6000-bench/blob/main/bench/deepseek-v4.1-flash_TP4_sglang/L16-REAL-TRAFFIC-2026-09-25.md
- github.com/ambientlight/rtx-pro-6000-bench
- github.com/ambientlight/rtx-pro-6000-bench/tree/main/docker/deepseek-v41
Related recipes
8× DGX Spark
74tok/sEveryday
1 request, generation at 32K context (speed table, ring of eight), thinking on, high, with DSpark (per recipe); experimental profile Source
Intelligence40with thinkingno thinking 25
- Custom kernel required
- Experimental
4× DGX Spark
71tok/sEveryday
1 request, decode at 32K context (speed table, ring of four), thinking on, high Source
Intelligence40with thinkingno thinking 25
- Custom kernel required
4× DGX Spark
59.1tok/sEveryday
1 request, prose prompt, with DSpark (per recipe) Source
Intelligence40with thinkingno thinking 25
3× DGX Spark
51tok/sEveryday
1 request, prose prompt, with MTP/EAGLE (spekulativ, ~59 ms/step) (per recipe) Source
Intelligence40with thinkingno thinking 25
- Custom kernel required
2× DGX Spark
42.6tok/sEveryday
1 request, prose prompt, with DSpark (per recipe) Source
Intelligence40with thinkingno thinking 25Original model's value. Heavily compressed (~2 bit), intelligence is likely much lower.
- ≤ 3 bit: quantization may cost quality
- Experimental
- 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.