Model benchmarks
C:\AI Models\unsloth\Qwen3.8-27B-GGUF\Qwen3.8-27B-UD-Q3_K_XL.gguf local LLM performance
As of September 2026, C:\AI Models\unsloth\Qwen3.8-27B-GGUF\Qwen3.8-27B-UD-Q3_K_XL.gguf runs at up to 68.4 tok/s for local inference (best of 3 community benchmark runs across 1 GPU).
Model size
27B
Peak speed
68.4 tok/s
Average speed
67.2 tok/s
Min memory
13.2 GB
Max context
65,536 tokens
Avg output / run
94,026 tokens
Avg runtime / run
23m 4s
Avg quality
80.4
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 | 76.0 | 80.4 | 85.5 |
| Agent Workflow | 86.1 | 87.1 | 88.5 |
| Code Generation | 37.6 | 55.7 | 73.0 |
| Role Play & Narrative | 89.6 | 92.4 | 95.4 |
| Research & Analysis | 85.2 | 86.5 | 87.8 |
Performance by hardware and tool
Every hardware/tool/quantization combination C:\AI Models\unsloth\Qwen3.8-27B-GGUF\Qwen3.8-27B-UD-Q3_K_XL.gguf has been benchmarked on, ranked by peak token generation speed. Last updated September 2026.
| Hardware | Tool | Quant | Best tok/s | Avg tok/s | Memory | Context | Quality | Runs |
|---|---|---|---|---|---|---|---|---|
| NVIDIA GeForce RTX 5070 Ti | llama.cpp | Q3_K | 68.4 tok/s | 67.2 tok/s | 13.2 GB | 65,536 tokens | 80.4 | 3 |
Frequently asked questions
- Is C:\AI Models\unsloth\Qwen3.8-27B-GGUF\Qwen3.8-27B-UD-Q3_K_XL.gguf good for coding?
- In our benchmarks, C:\AI Models\unsloth\Qwen3.8-27B-GGUF\Qwen3.8-27B-UD-Q3_K_XL.gguf scores 55.7/100 for coding. It runs at about 67.2 tok/s, so if you want more speed, Gemma4:E2B/QAT-MTP@131K is faster (~303.9 tok/s) and still scores well for coding (66.9/100).
- Is C:\AI Models\unsloth\Qwen3.8-27B-GGUF\Qwen3.8-27B-UD-Q3_K_XL.gguf good for agentic (tool-using) tasks?
- In our benchmarks, C:\AI Models\unsloth\Qwen3.8-27B-GGUF\Qwen3.8-27B-UD-Q3_K_XL.gguf scores 87.1/100 for agentic workflows. It runs at about 67.2 tok/s, so if you want more speed, unsloth/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-GGUF:UD-Q4_K_XL is faster (~169.2 tok/s) and still scores well for agentic workflows (83.0/100).
- How fast is C:\AI Models\unsloth\Qwen3.8-27B-GGUF\Qwen3.8-27B-UD-Q3_K_XL.gguf for local inference?
- Across 3 community benchmark runs, C:\AI Models\unsloth\Qwen3.8-27B-GGUF\Qwen3.8-27B-UD-Q3_K_XL.gguf reaches up to 68.4 tok/s and averages 67.2 tok/s, with the fastest results on NVIDIA GeForce RTX 5070 Ti.
- How much memory does C:\AI Models\unsloth\Qwen3.8-27B-GGUF\Qwen3.8-27B-UD-Q3_K_XL.gguf need?
- The leanest observed configuration used about 13.2 GB of memory (quantizations tested: Q3_K).
- Which tools have been used to run C:\AI Models\unsloth\Qwen3.8-27B-GGUF\Qwen3.8-27B-UD-Q3_K_XL.gguf?
- Benchmarks were submitted using llama.cpp. Results are community-contributed and updated as new runs arrive.