Model benchmarks
Huihui-Qwen3.8-27B-abliterated-Swift-1.5-GSQ-RCO-IQ3_S-mtp local LLM performance
As of November 2026, Huihui-Qwen3.8-27B-abliterated-Swift-1.5-GSQ-RCO-IQ3_S-mtp runs at up to 46.5 tok/s for local inference (best of 2 community benchmark runs across 1 GPU).
Model size
27B
Peak speed
46.5 tok/s
Average speed
43.5 tok/s
Avg PP
350.3 tok/s
Min memory
16.3 GB
Max context
65,536 tokens
Avg output / run
88,576 tokens
Avg runtime / run
33m 24s
Avg quality
89.1
Benchmark runs
2
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 2 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 | 88.5 | 89.1 | 89.7 |
| Agent Workflow | 87.4 | 88.9 | 90.4 |
| Code Generation | 85.0 | 86.6 | 88.2 |
| Role Play & Narrative | 94.5 | 94.6 | 94.7 |
| Research & Analysis | 84.0 | 86.4 | 88.7 |
Performance by hardware and tool
Every hardware/tool/quantization combination Huihui-Qwen3.8-27B-abliterated-Swift-1.5-GSQ-RCO-IQ3_S-mtp 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 |
|---|---|---|---|---|---|---|---|---|
| AMD Radeon RX 7800 XT | llama.cpp | — | 46.5 tok/s | 43.5 tok/s | 16.3 GB | 65,536 tokens | 89.1 | 2 |
Benchmark runs
All 2 Huihui-Qwen3.8-27B-abliterated-Swift-1.5-GSQ-RCO-IQ3_S-mtp runs submitted so far — expand one for its hardware, quality breakdown and per-scenario detail.
Frequently asked questions
- Is Huihui-Qwen3.8-27B-abliterated-Swift-1.5-GSQ-RCO-IQ3_S-mtp good for coding?
- In our benchmarks, Huihui-Qwen3.8-27B-abliterated-Swift-1.5-GSQ-RCO-IQ3_S-mtp scores 86.6/100 for coding, currently the best-scoring model for coding that runs on consumer hardware (≤24 GB VRAM). It runs at about 43.5 tok/s, so if you want more speed, ornith-1.5-35b-a3b-apex-i-quality is faster (~73.2 tok/s) and still scores well for coding (83.1/100).
- Is Huihui-Qwen3.8-27B-abliterated-Swift-1.5-GSQ-RCO-IQ3_S-mtp good for agentic (tool-using) tasks?
- In our benchmarks, Huihui-Qwen3.8-27B-abliterated-Swift-1.5-GSQ-RCO-IQ3_S-mtp scores 88.9/100 for agentic workflows. It runs at about 43.5 tok/s, so if you want more speed, Nex-N2.5-mini-APEX-Mini is faster (~88.2 tok/s) and still scores well for agentic workflows (91.3/100).
- How fast is Huihui-Qwen3.8-27B-abliterated-Swift-1.5-GSQ-RCO-IQ3_S-mtp for local inference?
- Across 2 community benchmark runs, Huihui-Qwen3.8-27B-abliterated-Swift-1.5-GSQ-RCO-IQ3_S-mtp reaches up to 46.5 tok/s and averages 43.5 tok/s, with the fastest results on AMD Radeon RX 7800 XT.
- How much memory does Huihui-Qwen3.8-27B-abliterated-Swift-1.5-GSQ-RCO-IQ3_S-mtp need?
- The leanest observed configuration used about 16.3 GB of memory.
- Which tools have been used to run Huihui-Qwen3.8-27B-abliterated-Swift-1.5-GSQ-RCO-IQ3_S-mtp?
- Benchmarks were submitted using llama.cpp. Results are community-contributed and updated as new runs arrive.