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).

llama.cpp
ShareRedditX

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.

TaskP5 (low)AvgP95 (high)
Overall84.085.687.5
Agent Workflow76.182.292.1
Code Generation82.884.886.2
Role Play & Narrative88.190.692.5
Research & Analysis81.984.687.1

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.

HardwareToolQuantBest tok/sAvg tok/sMemoryContextQualityRuns
NVIDIA GeForce RTX 3080 Tillama.cpp—88.7 tok/s85.5 tok/s15.7 GB196,608 tokens85.63

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.