Qwen3.6-35B-A3B-Escha-W2 · SGLang · RTX 5090 32GB
English | 简体中文 | 日本語 · ← All recipes
Source level: 📣 Claimed · Date: 2026-07-22 · Numbers come from the source. This repo did not measure them again.
Hardware
- RTX 5090 32GB (NVIDIA consumer)
- Reference cost: ≈ $5,400
Model
- Qwen3.6-35B-A3B-Escha-W2 ·
Escha W2 (INT8=on) - Download: huggingface.co/EschaLabs/Qwen3.6-35B-A3B-Escha-W2
Engine
- SGLang — install: https://github.com/sgl-project/sglang
- Note: SGLang (sglang/serve.sh)
Launch command
MEM=0.78 CTXLEN=32768 bash sglang/serve.sh (on 5090 add INT8=on ATTN_BACKEND=triton MEM=0.82)
Key parameters
- Context: 32K (16GB card 8K×16)
- Concurrency: single stream 283; bs32 about 2,670 (peak 3,017.6 @32 streams)
- Tags:
2bit,8bit,MoE
Expected speed
- 283 tok/s decode (📣 Claimed)
Caveats
- Not a first-hand measurement (claimed or relayed). Your results can differ.
- Speed changes with drivers, engine version, context length, and prompt length.
Source & attribution
- https://huggingface.co/EschaLabs/Qwen3.6-35B-A3B-Escha-W2
- Credit to the original author. This page only summarizes and links. Data under CC BY 4.0.
Source file: recipes/hf-eschalabs-qwen3-6-35b-a3b-escha-w2-1.md