4× RTX PRO 6000
139tok/sEveryday
1 request, prose prompt, with MTP (per recipe) Source
Intelligence18with thinkingno thinking 21
- Custom kernel required
Local AI · Recipe · 8× RTX 3090
Qwen3.5 35B-A3B with vLLM on 8× RTX 3090
Qwen3.5 35B-A3B with vLLM on 8× RTX 3090: 13.9 tok/s according to github.com (dataset as of Nov 3, 2026).
by 0xSero
Intelligence (original model) · Artificial Analysis19with thinkingno thinking 15
Engine
- Engine
- vLLM
- Quantization
- —
- Model family
- Qwen3.5
- Context
- —
- Parameters
- 35B MoE / 3B active
- Creator
- 0xSero
- GitHub stars
- 151
- Repo updated
- Sep 10, 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
Time to first token
Shorter is better.
What you need
- Hardware
- 8 × NVIDIA RTX 3090
- Weights
- Main weightsQwen/Qwen3.5-35B-A3B Apache-2.0
- Engine
- vLLM
Notes
What matters before you rebuild it.
- Experimental
- Offload setup
- Research
Sources
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.