Model benchmarks
qwen3.8-27b local LLM performance
As of August 2026, qwen3.8-27b runs at up to 51.8 tok/s for local inference (best of 2 community benchmark runs across 2 GPUs).
LM Studio
Model size
27B
Peak speed
51.8 tok/s
Average speed
32.5 tok/s
Min memory
16.5 GB
Max context
8,192 tokens
Best quality
93.0
Benchmark runs
2
GPUs tested
2
Performance by hardware and tool
Every hardware/tool/quantization combination qwen3.8-27b has been benchmarked on, ranked by peak token generation speed. Last updated August 2026.
| Hardware | Tool | Quant | Best tok/s | Avg tok/s | Memory | Context | Quality | Runs |
|---|---|---|---|---|---|---|---|---|
| AMD Radeon RX 7900 XTX | LM Studio | — | 51.8 tok/s | 51.8 tok/s | 16.5 GB | 8,192 tokens | 51.5 | 1 |
| Apple M5 Pro | LM Studio | — | 13.3 tok/s | 13.3 tok/s | 17.6 GB | 8,192 tokens | 44.6 | 1 |
Frequently asked questions
- How fast is qwen3.8-27b for local inference?
- Across 2 community benchmark runs, qwen3.8-27b reaches up to 51.8 tok/s and averages 32.5 tok/s, with the fastest results on AMD Radeon RX 7900 XTX.
- How much memory does qwen3.8-27b need?
- The leanest observed configuration used about 16.5 GB of memory.
- Which tools have been used to run qwen3.8-27b?
- Benchmarks were submitted using LM Studio. Results are community-contributed and updated as new runs arrive.