community pool / model answer

Qwen3.5-4B-quantized.w4a16

chat · RedHatAI/Qwen3.5-4B-quantized.w4a16

191.7 tok/s · model median · 3 runs in the model record.

Answer

236.9tok/s

reportedArc Pro B70 32GB ×24-bit · 1 run · decodebest cell in the pool

  1. Arc Pro B70 32GB ×2236.9 tok/sreported1 run · peak VRAM no data yet · 64 GB rig memory
  2. Arc Pro B70 32GB184.5 tok/sreported2 runs · peak VRAM no data yet · 32 GB rig memory
Reference rigQuantizationMedian resultBasisTTFT (ms)Peak VRAMMax contextEngines
Arc Pro B70 32GB4-bit184.5 tok/sreported44- GB1,024vllm
Arc Pro B70 32GB ×24-bit236.9 tok/sreported39- GB1,024vllm

Context tested

The table above is the tested context for each exact rig and quantization cell. Empty fields remain empty rather than being filled with an estimate.