Model benchmarks
qwen/qwen3.5-35b-a3b local LLM performance
As of April 2026, qwen/qwen3.5-35b-a3b runs at up to 32.0 tok/s for local inference (best of 2 community benchmark runs across 1 GPU).
Model size
35B
Peak speed
32.0 tok/s
Average speed
30.6 tok/s
Min memory
17.1 GB
Max context
16.384 tokens
Avg quality
71.7
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 | 71.7 | 71.7 | 71.8 |
| Agent Workflow | 66.2 | 70.9 | 75.6 |
| Code Generation | 32.8 | 47.5 | 62.2 |
| Role Play & Narrative | 78.3 | 84.9 | 91.5 |
| Research & Analysis | 80.5 | 83.7 | 86.8 |
Performance by hardware and tool
Every hardware/tool/quantization combination qwen/qwen3.5-35b-a3b has been benchmarked on, ranked by peak token generation speed. Last updated April 2026.
| Hardware | Tool | Quant | Best tok/s | Avg tok/s | Memory | Context | Quality | Runs |
|---|---|---|---|---|---|---|---|---|
| AMD Radeon RX 7900 XTX | LM Studio | — | 32.0 tok/s | 30.6 tok/s | 17.1 GB | 16.384 tokens | 71.7 | 2 |
Benchmark runs
All 2 qwen/qwen3.5-35b-a3b runs submitted so far — expand one for its hardware, quality breakdown and per-scenario detail.
Frequently asked questions
- Is qwen/qwen3.5-35b-a3b good for coding?
- In our benchmarks, qwen/qwen3.5-35b-a3b scores 47.5/100 for coding. It runs at about 30.6 tok/s, so if you want more speed, /home/polaris/models/Tiel-Coder-35B-A3B-MTP-UD-Q6_K_XL.gguf is faster (~35.6 tok/s) and still scores well for coding (84.9/100).
- Is qwen/qwen3.5-35b-a3b good for agentic (tool-using) tasks?
- In our benchmarks, qwen/qwen3.5-35b-a3b scores 70.9/100 for agentic workflows. It runs at about 30.6 tok/s, so if you want more speed, granite-4.2-8b-Q4_K_L is faster (~71.6 tok/s) and still scores well for agentic workflows (92.9/100).
- How fast is qwen/qwen3.5-35b-a3b for local inference?
- Across 2 community benchmark runs, qwen/qwen3.5-35b-a3b reaches up to 32.0 tok/s and averages 30.6 tok/s, with the fastest results on AMD Radeon RX 7900 XTX.
- How much memory does qwen/qwen3.5-35b-a3b need?
- The leanest observed configuration used about 17.1 GB of memory.
- Which tools have been used to run qwen/qwen3.5-35b-a3b?
- Benchmarks were submitted using LM Studio. Results are community-contributed and updated as new runs arrive.