Meta17B active / 400B totalQ4_K_M 下约 242 GB 显存

Llama 4 Maverick 17B — 显存、速度与本地部署

作者: Jakub Rusinowski · 最后更新:

The premium Llama 4 model. 128 experts with 17B active parameters. Matches GPT-4o in benchmarks. Its 400B total parameters must all be resident, which is roughly 242 GB at Q4_K_M — a multi-GPU server or a 256 GB+ unified-memory system. The 17B active count sets its speed, not its memory footprint.

Llama 4 Maverick 17B 在 Q4_K_M 下约需 242 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。

逻辑94
创意90
编程92

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。

量化显存速度(估算)适配
Q2_K
2.63 bpw
133.9 GB—放不下
Q3_K_M
3.41 bpw
172.9 GB—放不下
Q4_K_M
4.83 bpw
243.9 GB—放不下
Q5_K_M
5.67 bpw
285.9 GB—放不下
Q6_K
6.56 bpw
330.4 GB—放不下
Q8_0
8.50 bpw
427.4 GB—放不下
F16
16.00 bpw
802.4 GB—放不下

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. Llama 4 Maverick 17B 显存计算器 →

运行 Llama 4 Maverick 17B

目录中能运行 Llama 4 Maverick 17B 的最便宜 GPU 是 Apple M3 Ultra (512 GB).

购买此硬件 Apple Mac Studio M3 Ultra — 512 GB VRAM · 60 W board power立即云端部署 RunPod

或在 Vast.ai 比较

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。

联盟营销声明: 本页部分链接为联盟推广链接——如果你通过它们购买,LLM Configurator 可能会获得佣金,而你无需支付任何额外费用。作为亚马逊联盟成员(Amazon Associate),LLM Configurator 会从符合条件的购买中获得收益。
Apple Mac Studio M3 Ultra
512 GB VRAM · 60 W board power
2026年价格波动较大——请以当前商品页价格为准。

如何运行 Llama 4 Maverick 17B

安装 Ollama,然后运行:

ollama run llama4:maverick

规格

Curated — A hand-written entry from before this catalogue recorded its sources. The figures are long-standing but their provenance is not on file.

参数量
400 Billion (17B active)
上下文窗口
1,000,000
架构
MoE (128 experts)
提供商
Meta
许可证
Llama 4 Community
Practical deployment
Workstation (large unified memory or pro GPU)
规格量化
Q4_K_M
系统内存
32 GB
记录更新于
2025-04-05
许可证Llama Community允许商业使用

Weights are downloadable and commercial use is permitted, subject to the licence’s acceptable-use terms.

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。

最适合codingreasoningchatcomplex tasks

我的 GPU 能运行 Llama 4 Maverick 17B 吗?

Llama 4 的其他尺寸

Llama 4 Maverick 17B — 常见问题

How much VRAM does Llama 4 Maverick 17B need?

About 242 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 Llama 4 Maverick 17B run on an RTX 4090 (24 GB)?

No. Llama 4 Maverick 17B needs about 242 GB at Q4_K_M, more than a single RTX 4090's 24 GB. It needs a larger card, several GPUs, or Apple Silicon with enough unified memory — or it runs with part of the weights offloaded to system RAM, which is much slower.

How do I run Llama 4 Maverick 17B locally?

Install Ollama and run `ollama run llama4:maverick`. That pulls the weights and starts a local OpenAI-compatible endpoint; after the download nothing leaves the machine.

What other sizes does Llama 4 come in?

Llama 4 Scout 17B (67 GB), Llama 4 Maverick 17B (242 GB). Every size shares the family's training and licence; the larger ones score higher and need proportionally more memory.