Model benchmarks
perplexity-ai--pplx-computer-qwen-3-6-35b-a3b-mlx-20260709 local LLM performance
As of September 2026, perplexity-ai--pplx-computer-qwen-3-6-35b-a3b-mlx-20260709 runs at up to 81.4 tok/s for local inference (best of 2 community benchmark runs across 1 GPU).
Model size
35B
Peak speed
81.4 tok/s
Average speed
80.9 tok/s
Avg PP
612.1 tok/s
Min memory
18.7 GB
Max context
65.536 tokens
Avg output / run
14.267 tokens
Avg runtime / run
3m 9s
Avg quality
75.6
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 | 75.5 | 75.6 | 75.8 |
| Agent Workflow | 67.4 | 76.2 | 84.9 |
| Code Generation | 58.3 | 59.3 | 60.2 |
| Role Play & Narrative | 75.9 | 84.9 | 93.9 |
| Research & Analysis | 81.9 | 82.3 | 82.8 |
Performance by hardware and tool
Every hardware/tool/quantization combination perplexity-ai--pplx-computer-qwen-3-6-35b-a3b-mlx-20260709 has been benchmarked on, ranked by peak token generation speed. Last updated September 2026.
| Hardware | Tool | Quant | Best tok/s | Avg tok/s | Memory | Context | Quality | Runs |
|---|---|---|---|---|---|---|---|---|
| Apple M4 Pro | oMLX | — | 81.4 tok/s | 80.9 tok/s | 18.7 GB | 65.536 tokens | 75.6 | 2 |
Benchmark runs
All 2 perplexity-ai--pplx-computer-qwen-3-6-35b-a3b-mlx-20260709 runs submitted so far — expand one for its hardware, quality breakdown and per-scenario detail.
Frequently asked questions
- Is perplexity-ai--pplx-computer-qwen-3-6-35b-a3b-mlx-20260709 good for coding?
- In our benchmarks, perplexity-ai--pplx-computer-qwen-3-6-35b-a3b-mlx-20260709 scores 59.3/100 for coding. It runs at about 80.9 tok/s, so if you want more speed, unsloth/Qwen3.8-27B-GGUF:IQ3_S is faster (~104.3 tok/s) and still scores well for coding (83.0/100).
- Is perplexity-ai--pplx-computer-qwen-3-6-35b-a3b-mlx-20260709 good for agentic (tool-using) tasks?
- In our benchmarks, perplexity-ai--pplx-computer-qwen-3-6-35b-a3b-mlx-20260709 scores 76.2/100 for agentic workflows. It runs at about 80.9 tok/s, so if you want more speed, Tiel-Coder-35B-A3B-Q4_K_S is faster (~165.0 tok/s) and still scores well for agentic workflows (89.3/100).
- How fast is perplexity-ai--pplx-computer-qwen-3-6-35b-a3b-mlx-20260709 for local inference?
- Across 2 community benchmark runs, perplexity-ai--pplx-computer-qwen-3-6-35b-a3b-mlx-20260709 reaches up to 81.4 tok/s and averages 80.9 tok/s, with the fastest results on Apple M4 Pro.
- How much memory does perplexity-ai--pplx-computer-qwen-3-6-35b-a3b-mlx-20260709 need?
- The leanest observed configuration used about 18.7 GB of memory.
- Which tools have been used to run perplexity-ai--pplx-computer-qwen-3-6-35b-a3b-mlx-20260709?
- Benchmarks were submitted using oMLX. Results are community-contributed and updated as new runs arrive.