Model benchmarks
/models/hub/models--unsloth--Qwen3.8-27B-GGUF/snapshots/f1bfb127c64f7072bdd2cad55f258b9c8b2910fe/Qwen3.8-27B-UD-Q6_K_XL.gguf local LLM performance
As of August 2026, /models/hub/models--unsloth--Qwen3.8-27B-GGUF/snapshots/f1bfb127c64f7072bdd2cad55f258b9c8b2910fe/Qwen3.8-27B-UD-Q6_K_XL.gguf runs at up to 90.1 tok/s for local inference (best of 4 community benchmark runs across 1 GPU).
Model size
27B
Peak speed
90.1 tok/s
Average speed
84.3 tok/s
Min memory
36.9 GB
Max context
250,000 tokens
Avg output / run
44,664 tokens
Avg runtime / run
8m 56s
Avg quality
79.5
Benchmark runs
4
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 4 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 | 65.3 | 79.5 | 88.3 |
| Agent Workflow | 49.6 | 76.9 | 89.4 |
| Code Generation | 63.6 | 75.9 | 85.3 |
| Role Play & Narrative | 74.7 | 84.4 | 92.3 |
| Research & Analysis | 73.4 | 80.6 | 86.1 |
Performance by hardware and tool
Every hardware/tool/quantization combination /models/hub/models--unsloth--Qwen3.8-27B-GGUF/snapshots/f1bfb127c64f7072bdd2cad55f258b9c8b2910fe/Qwen3.8-27B-UD-Q6_K_XL.gguf 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 |
|---|---|---|---|---|---|---|---|---|
| NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition | vLLM | Q6_K | 90.1 tok/s | 84.3 tok/s | 36.9 GB | 250,000 tokens | 79.5 | 4 |
Frequently asked questions
- Is /models/hub/models--unsloth--Qwen3.8-27B-GGUF/snapshots/f1bfb127c64f7072bdd2cad55f258b9c8b2910fe/Qwen3.8-27B-UD-Q6_K_XL.gguf good for coding?
- In our benchmarks, /models/hub/models--unsloth--Qwen3.8-27B-GGUF/snapshots/f1bfb127c64f7072bdd2cad55f258b9c8b2910fe/Qwen3.8-27B-UD-Q6_K_XL.gguf scores 75.9/100 for coding. It runs at about 84.3 tok/s, so if you want more speed, mudler/Ornith-1.5-35B-A3B-APEX-MTP-GGUF is faster (~190.9 tok/s) and still scores well for coding (73.4/100).
- Is /models/hub/models--unsloth--Qwen3.8-27B-GGUF/snapshots/f1bfb127c64f7072bdd2cad55f258b9c8b2910fe/Qwen3.8-27B-UD-Q6_K_XL.gguf good for agentic (tool-using) tasks?
- In our benchmarks, /models/hub/models--unsloth--Qwen3.8-27B-GGUF/snapshots/f1bfb127c64f7072bdd2cad55f258b9c8b2910fe/Qwen3.8-27B-UD-Q6_K_XL.gguf scores 76.9/100 for agentic workflows. It runs at about 84.3 tok/s, so if you want more speed, Gemma4:E2B/QAT-MTP@131K is faster (~303.9 tok/s) and still scores well for agentic workflows (71.7/100).
- How fast is /models/hub/models--unsloth--Qwen3.8-27B-GGUF/snapshots/f1bfb127c64f7072bdd2cad55f258b9c8b2910fe/Qwen3.8-27B-UD-Q6_K_XL.gguf for local inference?
- Across 4 community benchmark runs, /models/hub/models--unsloth--Qwen3.8-27B-GGUF/snapshots/f1bfb127c64f7072bdd2cad55f258b9c8b2910fe/Qwen3.8-27B-UD-Q6_K_XL.gguf reaches up to 90.1 tok/s and averages 84.3 tok/s, with the fastest results on NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition.
- How much memory does /models/hub/models--unsloth--Qwen3.8-27B-GGUF/snapshots/f1bfb127c64f7072bdd2cad55f258b9c8b2910fe/Qwen3.8-27B-UD-Q6_K_XL.gguf need?
- The leanest observed configuration used about 36.9 GB of memory (quantizations tested: Q6_K).
- Which tools have been used to run /models/hub/models--unsloth--Qwen3.8-27B-GGUF/snapshots/f1bfb127c64f7072bdd2cad55f258b9c8b2910fe/Qwen3.8-27B-UD-Q6_K_XL.gguf?
- Benchmarks were submitted using vLLM. Results are community-contributed and updated as new runs arrive.