Model benchmarks
Qwen3.8-27B-GSQ-RCO-BuffedMod-IQ3_S_XL-mtp-Q6K local LLM performance
As of October 2026, Qwen3.8-27B-GSQ-RCO-BuffedMod-IQ3_S_XL-mtp-Q6K runs at up to 42.7 tok/s for local inference (best of 3 community benchmark runs across 2 GPUs).
Model size
27B
Peak speed
42.7 tok/s
Average speed
38.4 tok/s
Avg PP
433.5 tok/s
Min memory
16.6 GB
Max context
81,920 tokens
Avg output / run
54,378 tokens
Avg runtime / run
22m 32s
Avg quality
85.5
Benchmark runs
3
GPUs tested
2
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 | 82.3 | 85.5 | 87.7 |
| Agent Workflow | 81.4 | 83.8 | 86.8 |
| Code Generation | 73.2 | 79.6 | 85.8 |
| Role Play & Narrative | 85.7 | 90.6 | 94.2 |
| Research & Analysis | 87.5 | 87.9 | 88.6 |
Performance by hardware and tool
Every hardware/tool/quantization combination Qwen3.8-27B-GSQ-RCO-BuffedMod-IQ3_S_XL-mtp-Q6K has been benchmarked on, ranked by peak token generation speed. Last updated October 2026.
| Hardware | Tool | Quant | Best tok/s | Avg tok/s | Memory | Context | Quality | Runs |
|---|---|---|---|---|---|---|---|---|
| AMD Radeon RX 7800 XT | llama.cpp k8 v4 | — | 42.7 tok/s | 40.3 tok/s | 16.6 GB | 65,536 tokens | 87.3 | 2 |
| NVIDIA GeForce RTX 5060 | llama.cpp | — | 34.6 tok/s | 34.6 tok/s | 18.2 GB | 81,920 tokens | 81.8 | 1 |
Benchmark runs
All 3 Qwen3.8-27B-GSQ-RCO-BuffedMod-IQ3_S_XL-mtp-Q6K runs submitted so far — expand one for its hardware, quality breakdown and per-scenario detail.
Frequently asked questions
- Is Qwen3.8-27B-GSQ-RCO-BuffedMod-IQ3_S_XL-mtp-Q6K good for coding?
- In our benchmarks, Qwen3.8-27B-GSQ-RCO-BuffedMod-IQ3_S_XL-mtp-Q6K scores 79.6/100 for coding. It runs at about 38.4 tok/s, so if you want more speed, nvidia/Qwen3.8-27B-NVFP4 is faster (~75.5 tok/s) and still scores well for coding (83.5/100).
- Is Qwen3.8-27B-GSQ-RCO-BuffedMod-IQ3_S_XL-mtp-Q6K good for agentic (tool-using) tasks?
- In our benchmarks, Qwen3.8-27B-GSQ-RCO-BuffedMod-IQ3_S_XL-mtp-Q6K scores 83.8/100 for agentic workflows. It runs at about 38.4 tok/s, so if you want more speed, Nex-N2.5-mini-APEX-Mini is faster (~88.2 tok/s) and still scores well for agentic workflows (91.3/100).
- How fast is Qwen3.8-27B-GSQ-RCO-BuffedMod-IQ3_S_XL-mtp-Q6K for local inference?
- Across 3 community benchmark runs, Qwen3.8-27B-GSQ-RCO-BuffedMod-IQ3_S_XL-mtp-Q6K reaches up to 42.7 tok/s and averages 38.4 tok/s, with the fastest results on AMD Radeon RX 7800 XT.
- How much memory does Qwen3.8-27B-GSQ-RCO-BuffedMod-IQ3_S_XL-mtp-Q6K need?
- The leanest observed configuration used about 16.6 GB of memory.
- Which tools have been used to run Qwen3.8-27B-GSQ-RCO-BuffedMod-IQ3_S_XL-mtp-Q6K?
- Benchmarks were submitted using llama.cpp, llama.cpp k8 v4. Results are community-contributed and updated as new runs arrive.