Model benchmarks
qwen3.8-27b-Swift1.5-Uncensored-ATX-AI3 local LLM performance
As of November 2026, qwen3.8-27b-Swift1.5-Uncensored-ATX-AI3 runs at up to 88.7 tok/s for local inference (best of 3 community benchmark runs across 1 GPU).
Model size
27B
Peak speed
88.7 tok/s
Average speed
85.5 tok/s
Avg PP
759.3 tok/s
Min memory
15.7 GB
Max context
196,608 tokens
Avg output / run
42,097 tokens
Avg runtime / run
8m 4s
Avg quality
85.6
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 | 84.0 | 85.6 | 87.5 |
| Agent Workflow | 76.1 | 82.2 | 92.1 |
| Code Generation | 82.8 | 84.8 | 86.2 |
| Role Play & Narrative | 88.1 | 90.6 | 92.5 |
| Research & Analysis | 81.9 | 84.6 | 87.1 |
Performance by hardware and tool
Every hardware/tool/quantization combination qwen3.8-27b-Swift1.5-Uncensored-ATX-AI3 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 |
|---|---|---|---|---|---|---|---|---|
| NVIDIA GeForce RTX 3080 Ti | llama.cpp | — | 88.7 tok/s | 85.5 tok/s | 15.7 GB | 196,608 tokens | 85.6 | 3 |
Benchmark runs
All 3 qwen3.8-27b-Swift1.5-Uncensored-ATX-AI3 runs submitted so far — expand one for its hardware, quality breakdown and per-scenario detail.
Frequently asked questions
- Is qwen3.8-27b-Swift1.5-Uncensored-ATX-AI3 good for coding?
- In our benchmarks, qwen3.8-27b-Swift1.5-Uncensored-ATX-AI3 scores 84.8/100 for coding, among the top 2 for coding on consumer hardware (≤24 GB VRAM). It runs at about 85.5 tok/s, so if you want more speed, ornith-1.5-35b-a3b-apex-mtp-i-quality is faster (~105.2 tok/s) and still scores well for coding (83.6/100).
- Is qwen3.8-27b-Swift1.5-Uncensored-ATX-AI3 good for agentic (tool-using) tasks?
- In our benchmarks, qwen3.8-27b-Swift1.5-Uncensored-ATX-AI3 scores 82.2/100 for agentic workflows. It runs at about 85.5 tok/s, so if you want more speed, ornith-1.5-35b-a3b-apex-mtp-quality is faster (~103.2 tok/s) and still scores well for agentic workflows (88.5/100).
- How fast is qwen3.8-27b-Swift1.5-Uncensored-ATX-AI3 for local inference?
- Across 3 community benchmark runs, qwen3.8-27b-Swift1.5-Uncensored-ATX-AI3 reaches up to 88.7 tok/s and averages 85.5 tok/s, with the fastest results on NVIDIA GeForce RTX 3080 Ti.
- How much memory does qwen3.8-27b-Swift1.5-Uncensored-ATX-AI3 need?
- The leanest observed configuration used about 15.7 GB of memory.
- Which tools have been used to run qwen3.8-27b-Swift1.5-Uncensored-ATX-AI3?
- Benchmarks were submitted using llama.cpp. Results are community-contributed and updated as new runs arrive.