Model benchmarks
glm-4.7-flash:latest local LLM performance
As of September 2026, glm-4.7-flash:latest runs at up to 90.5 tok/s for local inference (best of 4 community benchmark runs across 2 GPUs).
Model size
29.9B
Peak speed
90.5 tok/s
Average speed
38.8 tok/s
Avg PP
676.4 tok/s
Min memory
24.3 GB
Max context
65,536 tokens
Avg output / run
18,511 tokens
Avg runtime / run
5m 39s
Avg quality
54.6
Benchmark runs
4
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 4 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 | 23.7 | 54.6 | 69.2 |
| Agent Workflow | 16.9 | 56.8 | 79.9 |
| Code Generation | 8.1 | 45.5 | 64.8 |
| Role Play & Narrative | 56.6 | 67.6 | 84.9 |
| Research & Analysis | 10.8 | 48.3 | 70.1 |
Performance by hardware and tool
Every hardware/tool/quantization combination glm-4.7-flash:latest 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 |
|---|---|---|---|---|---|---|---|---|
| CPU only | cpa | — | 90.5 tok/s | 90.5 tok/s | n/a | 8,192 tokens | 69.2 | 1 |
| Apple M4 Max | LM Studio | — | 50.7 tok/s | 50.7 tok/s | n/a | 8,192 tokens | 16.7 | 1 |
| AMD Radeon RX 7900 XTX | Ollama | Q4_K_M | 7.9 tok/s | 7.0 tok/s | 24.3 GB | 65,536 tokens | 66.2 | 2 |
Benchmark runs
All 4 glm-4.7-flash:latest runs submitted so far — expand one for its hardware, quality breakdown and per-scenario detail.
Frequently asked questions
- Is glm-4.7-flash:latest good for coding?
- In our benchmarks, glm-4.7-flash:latest scores 45.5/100 for coding. It runs at about 38.8 tok/s, so if you want more speed, local-qwen is faster (~60.2 tok/s) and still scores well for coding (84.8/100).
- Is glm-4.7-flash:latest good for agentic (tool-using) tasks?
- In our benchmarks, glm-4.7-flash:latest scores 56.8/100 for agentic workflows. It runs at about 38.8 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 glm-4.7-flash:latest for local inference?
- Across 4 community benchmark runs, glm-4.7-flash:latest reaches up to 90.5 tok/s and averages 38.8 tok/s.
- How much memory does glm-4.7-flash:latest need?
- The leanest observed configuration used about 24.3 GB of memory (quantizations tested: Q4_K_M).
- Which tools have been used to run glm-4.7-flash:latest?
- Benchmarks were submitted using LM Studio, Ollama, cpa. Results are community-contributed and updated as new runs arrive.