Model benchmarks
unsloth/Qwen3.8-27B-GGUF:IQ3_S local LLM performance
As of August 2026, unsloth/Qwen3.8-27B-GGUF:IQ3_S runs at up to 110.7 tok/s for local inference (best of 5 community benchmark runs across 1 GPU).
Model size
27B
Peak speed
110.7 tok/s
Average speed
104.3 tok/s
Min memory
n/a
Max context
131,072 tokens
Avg output / run
84,712 tokens
Avg runtime / run
13m 11s
Avg quality
86.8
Benchmark runs
5
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 5 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 | 86.2 | 86.8 | 88.0 |
| Agent Workflow | 82.0 | 86.2 | 90.4 |
| Code Generation | 80.3 | 83.0 | 86.5 |
| Role Play & Narrative | 87.8 | 90.0 | 93.4 |
| Research & Analysis | 87.6 | 88.1 | 88.7 |
Performance by hardware and tool
Every hardware/tool/quantization combination unsloth/Qwen3.8-27B-GGUF:IQ3_S 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 GeForce RTX 4090 | llama.cpp | — | 110.7 tok/s | 104.3 tok/s | n/a | 131,072 tokens | 86.8 | 5 |
Benchmark runs
All 5 unsloth/Qwen3.8-27B-GGUF:IQ3_S runs submitted so far — expand one for its hardware, quality breakdown and per-scenario detail.
Frequently asked questions
- Is unsloth/Qwen3.8-27B-GGUF:IQ3_S good for coding?
- In our benchmarks, unsloth/Qwen3.8-27B-GGUF:IQ3_S scores 83.0/100 for coding. It runs at about 104.3 tok/s, so if you want more speed, Qwen3.6-35B-A3B-Q4_K_XL is faster (~156.9 tok/s) and still scores well for coding (82.2/100).
- Is unsloth/Qwen3.8-27B-GGUF:IQ3_S good for agentic (tool-using) tasks?
- In our benchmarks, unsloth/Qwen3.8-27B-GGUF:IQ3_S scores 86.2/100 for agentic workflows. It runs at about 104.3 tok/s, so if you want more speed, Tiel-Coder-35B-A3B-Q4_K_S is faster (~162.4 tok/s) and still scores well for agentic workflows (87.8/100).
- How fast is unsloth/Qwen3.8-27B-GGUF:IQ3_S for local inference?
- Across 5 community benchmark runs, unsloth/Qwen3.8-27B-GGUF:IQ3_S reaches up to 110.7 tok/s and averages 104.3 tok/s, with the fastest results on NVIDIA GeForce RTX 4090.
- Which tools have been used to run unsloth/Qwen3.8-27B-GGUF:IQ3_S?
- Benchmarks were submitted using llama.cpp. Results are community-contributed and updated as new runs arrive.