Model benchmarks
huihui-qwen3.6-35b-a3b-abliterated-q5_k local LLM performance
As of November 2026, huihui-qwen3.6-35b-a3b-abliterated-q5_k runs at up to 138.1 tok/s for local inference (best of 2 community benchmark runs across 1 GPU).
Model size
35B
Peak speed
138.1 tok/s
Average speed
136.4 tok/s
Avg PP
399.0 tok/s
Min memory
11.9 GB
Max context
65,536 tokens
Avg output / run
21,391 tokens
Avg runtime / run
2m 45s
Avg quality
74.1
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 | 67.5 | 74.1 | 80.6 |
| Agent Workflow | 40.2 | 57.4 | 74.5 |
| Code Generation | 75.6 | 78.8 | 82.0 |
| Role Play & Narrative | 83.3 | 85.2 | 87.1 |
| Research & Analysis | 71.0 | 74.9 | 78.7 |
Performance by hardware and tool
Every hardware/tool/quantization combination huihui-qwen3.6-35b-a3b-abliterated-q5_k 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 4070 | llama.cpp | Q5_K | 138.1 tok/s | 136.4 tok/s | 11.9 GB | 65,536 tokens | 74.1 | 2 |
Benchmark runs
All 2 huihui-qwen3.6-35b-a3b-abliterated-q5_k runs submitted so far — expand one for its hardware, quality breakdown and per-scenario detail.
Frequently asked questions
- Is huihui-qwen3.6-35b-a3b-abliterated-q5_k good for agentic (tool-using) tasks?
- In our benchmarks, huihui-qwen3.6-35b-a3b-abliterated-q5_k scores 57.4/100 for agentic workflows. It runs at about 136.4 tok/s, so if you want more speed, unsloth/Qwen3.6-35B-A3B-GGUF:UD-Q4_K_M is faster (~246.8 tok/s) and still scores well for agentic workflows (80.6/100).
- How fast is huihui-qwen3.6-35b-a3b-abliterated-q5_k for local inference?
- Across 2 community benchmark runs, huihui-qwen3.6-35b-a3b-abliterated-q5_k reaches up to 138.1 tok/s and averages 136.4 tok/s, with the fastest results on NVIDIA GeForce RTX 4070.
- How much memory does huihui-qwen3.6-35b-a3b-abliterated-q5_k need?
- The leanest observed configuration used about 11.9 GB of memory (quantizations tested: Q5_K).
- Which tools have been used to run huihui-qwen3.6-35b-a3b-abliterated-q5_k?
- Benchmarks were submitted using llama.cpp. Results are community-contributed and updated as new runs arrive.