Model benchmarks
unsloth/GLM-4.7-Flash-GGUF:UD-Q4_K_XL local LLM performance
As of October 2026, unsloth/GLM-4.7-Flash-GGUF:UD-Q4_K_XL runs at up to 117.7 tok/s for local inference (best of 3 community benchmark runs across 2 GPUs).
Model size
30B
Peak speed
117.7 tok/s
Average speed
113.7 tok/s
Avg PP
818.3 tok/s
Min memory
18.2 GB
Max context
131,072 tokens
Avg output / run
16,319 tokens
Avg runtime / run
2m 44s
Avg quality
65.3
Benchmark runs
3
GPUs tested
2
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 3 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 | 62.8 | 65.3 | 67.2 |
| Agent Workflow | 60.2 | 69.9 | 79.5 |
| Code Generation | 57.7 | 59.7 | 61.5 |
| Role Play & Narrative | 78.1 | 82.6 | 87.2 |
| Research & Analysis | 46.8 | 49.0 | 50.7 |
Performance by hardware and tool
Every hardware/tool/quantization combination unsloth/GLM-4.7-Flash-GGUF:UD-Q4_K_XL has been benchmarked on, ranked by peak token generation speed. Last updated October 2026.
| Hardware | Tool | Quant | Best tok/s | Avg tok/s | Memory | Context | Quality | Runs |
|---|---|---|---|---|---|---|---|---|
| AMD Radeon AI PRO R9700 | llama.cpp | Q4_K | 117.7 tok/s | 117.7 tok/s | 18.2 GB | 65,536 tokens | 67.3 | 1 |
| Advanced Micro Devices, Inc. [AMD/ATI] Device 7551 | llama.cpp | Q4_K | 112.3 tok/s | 111.6 tok/s | n/a | 131,072 tokens | 64.3 | 2 |
Benchmark runs
All 3 unsloth/GLM-4.7-Flash-GGUF:UD-Q4_K_XL runs submitted so far — expand one for its hardware, quality breakdown and per-scenario detail.
Frequently asked questions
- Is unsloth/GLM-4.7-Flash-GGUF:UD-Q4_K_XL good for agentic (tool-using) tasks?
- In our benchmarks, unsloth/GLM-4.7-Flash-GGUF:UD-Q4_K_XL scores 69.9/100 for agentic workflows. It runs at about 113.7 tok/s, so if you want more speed, Tiel-Coder-35B-A3B-MLX-oQ4e-MTP is faster (~151.8 tok/s) and still scores well for agentic workflows (83.0/100).
- How fast is unsloth/GLM-4.7-Flash-GGUF:UD-Q4_K_XL for local inference?
- Across 3 community benchmark runs, unsloth/GLM-4.7-Flash-GGUF:UD-Q4_K_XL reaches up to 117.7 tok/s and averages 113.7 tok/s, with the fastest results on AMD Radeon AI PRO R9700.
- How much memory does unsloth/GLM-4.7-Flash-GGUF:UD-Q4_K_XL need?
- The leanest observed configuration used about 18.2 GB of memory (quantizations tested: Q4_K).
- Which tools have been used to run unsloth/GLM-4.7-Flash-GGUF:UD-Q4_K_XL?
- Benchmarks were submitted using llama.cpp. Results are community-contributed and updated as new runs arrive.