community pool / model answer
Qwen3.6-27B-int4-AutoRound
chat · Lorbus/Qwen3.6-27B-int4-AutoRound
66.4 tok/s · model median · 21 runs in the model record.
| Reference rig | Quantization | Median result | Basis | TTFT (ms) | Peak VRAM | Max context | Engines |
|---|---|---|---|---|---|---|---|
| Arc Pro B70 32GB | 4-bit | 41.3 tok/s | measured | - | - GB | 4,096 | vllm |
| Arc Pro B70 32GB ×2 | 4-bit | 48.3 tok/s | measured | - | - GB | 4,096 | vllm |
| RTX 3090 24GB | 4-bit | 74.5 tok/s | measured | 480 | 22.5 GB | 32,768 | vllm |
| RTX 3090 24GB ×2 | 4-bit | 78.2 tok/s | measured | 79 | 22.8 GB | 262,144 | vllm |
| RTX 4090 24GB | 16-bit | 52.6 tok/s | reported | 88 | - GB | 2,048 | vllm |
| RTX 5060 Ti 16GB ×2 | 4-bit | 69.3 tok/s | reported | 94 | - GB | 2,048 | vllm |
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.