Clef — Cloudflare 的本地 AI 模型
作者： Jakub Rusinowski · 最后更新：
Cloudflare 的决策模型 Clef 27B 与 Clef-Flash 9B：输入带类型的问题，一次评分即输出每个选项的概率，不生成文本。两者都在 Ollama 库中，并支持图像输入。
变体
Clef 最小的变体在 Q4_K_M 下约需 6 GB 显存——量化权重加框架开销，不含 KV 缓存。
| 模型 | Q4 下显存 | 显存 | 上下文 | 运行 |
|---|---|---|---|---|
| Clef 27B → 27B | ~17.3 GB | 64K (documented) | ollama pull clef | |
| Clef-Flash 9B → 9B | ~6.5 GB | 64K (documented) | ollama pull clef-flash |
显存为 Q4_K_M 下的量化权重加开销，与 GPU 与显存检测器使用同一引擎计算。
如何在本地运行 Clef
安装 Ollama，然后拉取标签。
ollama pull clef在上方选择一个尺寸，查看它自己的显存、速度估算和安装命令。
许可证
Commercial use permitted. No usage restrictions beyond attribution.
适用于: Clef 27B, Clef-Flash 9B推荐 GPU
目录中能在本地运行 Clef（至少 6 GB 显存）的最便宜 GPU 是 Intel Arc B570 (10 GB).
Clef — 常见问题
How much VRAM does Clef need?
Clef needs about 6 GB VRAM at Q4_K_M quantization for its smallest variant. Variants: Clef 27B (17 GB, Q4_K_M); Clef-Flash 9B (6 GB, Q4_K_M). On Apple Silicon, unified memory counts toward this requirement.
Can I run Clef on an RTX 4090 (24 GB)?
Yes — Clef runs on an RTX 4090 (24 GB) and other 24 GB cards such as the RTX 3090. Smaller variants also fit comfortably on 8–16 GB GPUs at Q4_K_M.
What quantization should I use for Clef?
Q4_K_M is the best balance of quality and VRAM for Clef in most cases. Choose Q8_0 for near-lossless quality if you have spare VRAM, or smaller quants (Q3/Q2) only when memory is tight.
How do I run Clef locally?
Clef is a decision model: it is called over an HTTP endpoint (/v1/systemone), not chatted with. Where a variant is available in Ollama 0.35 or newer, pull it with `ollama pull` and send requests to the local server; the others ship their own server. Each variant page shows the exact commands for that model.