Model benchmarks
qwen3.8-35b-a3b-distill-miniplus-v2.1-abliterated local LLM performance
As of November 2026, qwen3.8-35b-a3b-distill-miniplus-v2.1-abliterated runs at up to 142.3 tok/s for local inference (best of 2 community benchmark runs across 1 GPU).
Model size
35B
Peak speed
142.3 tok/s
Average speed
138.9 tok/s
Avg PP
468.2 tok/s
Min memory
11.1 GB
Max context
65,536 tokens
Avg output / run
12,491 tokens
Avg runtime / run
1m 33s
Avg quality
77.4
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 | 77.1 | 77.4 | 77.7 |
| Agent Workflow | 70.9 | 75.6 | 80.2 |
| Code Generation | 72.6 | 73.9 | 75.3 |
| Role Play & Narrative | 74.7 | 78.8 | 82.8 |
| Research & Analysis | 81.1 | 81.5 | 82.0 |
Performance by hardware and tool
Every hardware/tool/quantization combination qwen3.8-35b-a3b-distill-miniplus-v2.1-abliterated 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 | — | 142.3 tok/s | 138.9 tok/s | 11.1 GB | 65,536 tokens | 77.4 | 2 |
Benchmark runs
All 2 qwen3.8-35b-a3b-distill-miniplus-v2.1-abliterated runs submitted so far — expand one for its hardware, quality breakdown and per-scenario detail.
Frequently asked questions
- Is qwen3.8-35b-a3b-distill-miniplus-v2.1-abliterated good for agentic (tool-using) tasks?
- In our benchmarks, qwen3.8-35b-a3b-distill-miniplus-v2.1-abliterated scores 75.6/100 for agentic workflows. It runs at about 138.9 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 qwen3.8-35b-a3b-distill-miniplus-v2.1-abliterated for local inference?
- Across 2 community benchmark runs, qwen3.8-35b-a3b-distill-miniplus-v2.1-abliterated reaches up to 142.3 tok/s and averages 138.9 tok/s, with the fastest results on NVIDIA GeForce RTX 4070.
- How much memory does qwen3.8-35b-a3b-distill-miniplus-v2.1-abliterated need?
- The leanest observed configuration used about 11.1 GB of memory.
- Which tools have been used to run qwen3.8-35b-a3b-distill-miniplus-v2.1-abliterated?
- Benchmarks were submitted using llama.cpp. Results are community-contributed and updated as new runs arrive.