Model benchmarks
unsloth/Qwen3.5-9B-MTP-GGUF local LLM performance
As of September 2026, unsloth/Qwen3.5-9B-MTP-GGUF runs at up to 102.9 tok/s for local inference (best of 2 community benchmark runs across 1 GPU).
Model size
9B
Peak speed
102.9 tok/s
Average speed
99.3 tok/s
Avg PP
784.8 tok/s
Min memory
4.4 GB
Max context
262.144 tokens
Avg output / run
18.037 tokens
Avg runtime / run
3m 14s
Avg quality
69.0
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 | 66.8 | 69.0 | 71.3 |
| Agent Workflow | 74.3 | 75.8 | 77.2 |
| Code Generation | 35.3 | 47.4 | 59.5 |
| Role Play & Narrative | 75.6 | 79.0 | 82.4 |
| Research & Analysis | 69.0 | 73.9 | 78.9 |
Performance by hardware and tool
Every hardware/tool/quantization combination unsloth/Qwen3.5-9B-MTP-GGUF 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 |
|---|---|---|---|---|---|---|---|---|
| NVIDIA GeForce RTX 4070 | llama.cpp | — | 102.9 tok/s | 99.3 tok/s | 4.4 GB | 262.144 tokens | 69.0 | 2 |
Benchmark runs
All 2 unsloth/Qwen3.5-9B-MTP-GGUF runs submitted so far — expand one for its hardware, quality breakdown and per-scenario detail.
Frequently asked questions
- Is unsloth/Qwen3.5-9B-MTP-GGUF good for coding?
- In our benchmarks, unsloth/Qwen3.5-9B-MTP-GGUF scores 47.4/100 for coding. It runs at about 99.3 tok/s, so if you want more speed, Qwen3.6-35B-A3B-Q4_K_XL is faster (~156.9 tok/s) and still scores well for coding (82.2/100).
- Is unsloth/Qwen3.5-9B-MTP-GGUF good for agentic (tool-using) tasks?
- In our benchmarks, unsloth/Qwen3.5-9B-MTP-GGUF scores 75.8/100 for agentic workflows. It runs at about 99.3 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 unsloth/Qwen3.5-9B-MTP-GGUF for local inference?
- Across 2 community benchmark runs, unsloth/Qwen3.5-9B-MTP-GGUF reaches up to 102.9 tok/s and averages 99.3 tok/s, with the fastest results on NVIDIA GeForce RTX 4070.
- How much memory does unsloth/Qwen3.5-9B-MTP-GGUF need?
- The leanest observed configuration used about 4.4 GB of memory.
- Which tools have been used to run unsloth/Qwen3.5-9B-MTP-GGUF?
- Benchmarks were submitted using llama.cpp. Results are community-contributed and updated as new runs arrive.