证据: 基准测试结果由厂商自行报告
LFM3-VL 2B — 显存、速度与本地部署
作者： Jakub Rusinowski · 最后更新：
LFM3 的视觉语言成员：相同的 24 层三重混合骨干加图像编码器，总参数 1.9B，上下文 128K。提供官方 GGUF、MLX 和 ONNX 版本；llama.cpp 自 2026 年 10 月 27 日起可加载。
LFM3-VL 2B 在 Q4_K_M 下约需 2 GB 显存——量化权重加框架开销，不含 KV 缓存。在 Apple Silicon 上，这部分来自统一内存。
按量化级别的显存与速度
计算基准：NVIDIA RTX 4090 (24 GB)。包含 8K 上下文的 KV 缓存，因此数值高于上方的主数字。
| 量化 | 显存 | 显存 | 速度（估算） | 适配 |
|---|---|---|---|---|
| Q2_K 2.63 bpw | 1.5 GB | ~317 tok/s | 可运行 | |
| Q3_K_M 3.41 bpw | 1.7 GB | ~294 tok/s | 可运行 | |
| Q4_K_M 4.83 bpw | 2.1 GB | ~259 tok/s | 可运行 | |
| Q5_K_M 5.67 bpw | 2.3 GB | ~243 tok/s | 可运行 | |
| Q6_K 6.56 bpw | 2.5 GB | ~227 tok/s | 可运行 | |
| Q8_0 8.50 bpw | 2.9 GB | ~199 tok/s | 可运行 | |
| F16 16.00 bpw | 4.7 GB | ~135 tok/s | 可运行 |
黑色标记 = NVIDIA RTX 4090 (24 GB) 上的可用显存。 估算来自内存带宽屋顶线模型，详见 方法说明页. LFM3-VL 2B 显存计算器 →
运行 LFM3-VL 2B
目录中能运行 LFM3-VL 2B 的最便宜 GPU 是 Intel Arc B570 (10 GB).
或在 Vast.ai 比较，低至 $0.35/小时 (typical low · varies)
作为亚马逊联盟成员，我们从符合条件的购买中获得收入。云 GPU 链接为推荐链接——我们可能获得佣金，您无需额外付费。
如何运行 LFM3-VL 2B
No Ollama tag for the vision model yet. Use the official GGUF with llama.cpp, or the MLX builds on Apple Silicon.
规格
Verified — Checked against the primary source — the model card or the vendor spec page — and corroborated by a second independent source.
- 参数量
- 1.9 Billion
- 上下文窗口
- 128,000
- 架构
- Hybrid: 13 short-conv + 5 sliding-window + 6 global-attention layers, image encoder
- 提供商
- Liquid AI
- 许可证
- LFM Open License v1.0
- Practical deployment
- Edge device (phone, SBC)
- 规格量化
- Q4_K_M
- 系统内存
- 8 GB
- 记录更新于
- 2026-11-03
Weights are downloadable and commercial use is permitted, subject to the licence’s acceptable-use terms.
Commercial use is licensed only below $10M annual revenue; above that, a separate agreement.
限制与注意事项
- Commercial use is licensed only below US$10M annual revenue.
质量与使用场景
评分由模型作者或独立评测方发布——衡量质量而非吞吐量，并非我们实测。
Liquid LFM3 的其他尺寸
LFM3-VL 2B — 常见问题
How much VRAM does LFM3-VL 2B need?
About 2 GB at Q4_K_M — quantized weights plus framework overhead, before any KV cache. The cache grows with context length and is added on top; the table above folds it in. Apple Silicon counts unified memory toward the same figure.
Does LFM3-VL 2B run on an RTX 4090 (24 GB)?
Yes. LFM3-VL 2B needs about 2 GB at Q4_K_M, inside a 24 GB card, at an estimated 259 tokens/sec.
How do I run LFM3-VL 2B locally?
No Ollama tag for the vision model yet. Use the official GGUF with llama.cpp, or the MLX builds on Apple Silicon. Running the published tag would send your prompts to a hosted GPU rather than your own machine.
What other sizes does Liquid LFM3 come in?
LFM3 1.7B (2 GB), LFM3-VL 2B (2 GB). Every size shares the family's training and licence; the larger ones score higher and need proportionally more memory.