Model benchmarks
Qwen3.8-27B-GSQ-RCO-Intel-Arc-Tuned-MTP-3.2BPW local LLM performance
As of November 2026, Qwen3.8-27B-GSQ-RCO-Intel-Arc-Tuned-MTP-3.2BPW runs at up to 19.2 tok/s for local inference (best of 3 community benchmark runs across 1 GPU).
Model size
27B
Peak speed
19.2 tok/s
Average speed
18.9 tok/s
Avg PP
254.3 tok/s
Min memory
16.0 GB
Max context
131,072 tokens
Avg output / run
57,153 tokens
Avg runtime / run
51m 44s
Avg quality
84.3
Benchmark runs
3
GPUs tested
1
Quality by task
Average LLM-judged quality (0–100) with the run-to-run spread shown as a P5–P95 band, overall and for each benchmark task, across all 3 runs. The low and high columns show how much the judge’s score varies between runs, and need at least two runs to display.
| Task | P5 (low) | Avg | P95 (high) |
|---|---|---|---|
| Overall | 83.2 | 84.3 | 85.1 |
| Agent Workflow | 90.9 | 92.2 | 94.2 |
| Code Generation | 55.9 | 63.2 | 67.7 |
| Role Play & Narrative | 91.9 | 93.5 | 94.4 |
| Research & Analysis | 86.5 | 88.5 | 90.4 |
Performance by hardware and tool
Every hardware/tool/quantization combination Qwen3.8-27B-GSQ-RCO-Intel-Arc-Tuned-MTP-3.2BPW has been benchmarked on, ranked by peak token generation speed. Last updated November 2026.
| Hardware | Tool | Quant | Best tok/s | Avg tok/s | Memory | Context | Quality | Runs |
|---|---|---|---|---|---|---|---|---|
| Advanced Micro Devices, Inc. [AMD/ATI] HawkPoint1 | llama.cpp | — | 19.2 tok/s | 18.9 tok/s | 16.0 GB | 131,072 tokens | 84.3 | 3 |
Benchmark runs
All 3 Qwen3.8-27B-GSQ-RCO-Intel-Arc-Tuned-MTP-3.2BPW runs submitted so far — expand one for its hardware, quality breakdown and per-scenario detail.
Frequently asked questions
- Is Qwen3.8-27B-GSQ-RCO-Intel-Arc-Tuned-MTP-3.2BPW good for coding?
- In our benchmarks, Qwen3.8-27B-GSQ-RCO-Intel-Arc-Tuned-MTP-3.2BPW scores 63.2/100 for coding. It runs at about 18.9 tok/s, so if you want more speed, Huihui-Qwen3.8-27B-abliterated-Swift-1.5-GSQ-RCO-IQ3_S-mtp is faster (~43.5 tok/s) and still scores well for coding (86.6/100).
- Is Qwen3.8-27B-GSQ-RCO-Intel-Arc-Tuned-MTP-3.2BPW good for agentic (tool-using) tasks?
- In our benchmarks, Qwen3.8-27B-GSQ-RCO-Intel-Arc-Tuned-MTP-3.2BPW scores 92.2/100 for agentic workflows, currently the best-scoring model for agentic workflows that runs on consumer hardware (≤24 GB VRAM). It runs at about 18.9 tok/s, so if you want more speed, muse-glimmer:latest is faster (~33.3 tok/s) and still scores well for agentic workflows (91.6/100).
- How fast is Qwen3.8-27B-GSQ-RCO-Intel-Arc-Tuned-MTP-3.2BPW for local inference?
- Across 3 community benchmark runs, Qwen3.8-27B-GSQ-RCO-Intel-Arc-Tuned-MTP-3.2BPW reaches up to 19.2 tok/s and averages 18.9 tok/s, with the fastest results on Advanced Micro Devices, Inc. [AMD/ATI] HawkPoint1.
- How much memory does Qwen3.8-27B-GSQ-RCO-Intel-Arc-Tuned-MTP-3.2BPW need?
- The leanest observed configuration used about 16.0 GB of memory.
- Which tools have been used to run Qwen3.8-27B-GSQ-RCO-Intel-Arc-Tuned-MTP-3.2BPW?
- Benchmarks were submitted using llama.cpp. Results are community-contributed and updated as new runs arrive.