NVIDIA31.6B (3.5B active)Q4_K_M 下约 20 GB 显存

Nemotron-Cascade 2 30B-A3B — 显存、速度与本地部署

作者: Jakub Rusinowski · 最后更新:

31.6B parameters resident — about 19.9 GB at Q4_K_M, so a 24 GB card holds it — with 3.5B active per token, which is why it decodes far faster than its size suggests. 52 layers: 23 Mamba-2 state-space, 23 MoE, and 6 GQA attention. Each MoE layer carries 128 routed experts plus one shared, six routed activated per token. Gold-medal level at IMO 2025, IOI and the ICPC World Finals, reached through post-training rather than scale — and NVIDIA published that recipe. 1M-token context, NVIDIA Open Model License.

Nemotron-Cascade 2 30B-A3B 在 Q4_K_M 下约需 20 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90
创意78
编程87

按量化级别的显存与速度

计算基准:NVIDIA RTX 4090 (24 GB)。仅含权重与开销:该模型架构未公开,因此未计入 KV 缓存。

量化显存速度(估算)适配
Q2_K
2.63 bpw
11.2 GB~201 tok/s可运行
Q3_K_M
3.41 bpw
14.3 GB~185 tok/s可运行
Q4_K_M
4.83 bpw
19.9 GB~160 tok/s可运行
Q5_K_M
5.67 bpw
23.2 GB~149 tok/s勉强
Q6_K
6.56 bpw
26.7 GB~21 tok/s需卸载
Q8_0
8.50 bpw
34.4 GB~18 tok/s需卸载
F16
16.00 bpw
64 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. Nemotron-Cascade 2 30B-A3B 显存计算器 →

运行 Nemotron-Cascade 2 30B-A3B

目录中能运行 Nemotron-Cascade 2 30B-A3B 的最便宜 GPU 是 AMD Radeon RX 7900 XTX (24 GB).

购买此硬件 AMD Radeon RX 7900 XTX 24GB — 24 GB VRAM · 355 W board power立即云端部署 RunPod 上的 RTX 4090 — 低至 $0.34/小时 · 价格核实于 2026-07

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。

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。
AMD Radeon RX 7900 XTX 24GB
24 GB VRAM · 355 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 Nemotron-Cascade 2 30B-A3B

安装 Ollama,然后运行:

ollama run nemotron-cascade-2:30b
Hugging Face 上的权重: nvidia/Nemotron-Cascade-2-30B-A3B ↗

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
31.6 Billion (3.5B active)
上下文窗口
1,048,576
架构
Hybrid Mamba-2 + MoE + GQA attention
提供商
NVIDIA
许可证
NVIDIA Open Model License
Practical deployment
Single consumer GPU
规格量化
Q4_K_M
系统内存
32 GB
记录更新于
2026-09-11
许可证NVIDIA Open Model License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reasoningagentic tasksmathlong context

我的 GPU 能运行 Nemotron-Cascade 2 30B-A3B 吗?

Nemotron-Cascade 2 30B-A3B — 常见问题

How much VRAM does Nemotron-Cascade 2 30B-A3B need?

About 20 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 Nemotron-Cascade 2 30B-A3B run on an RTX 4090 (24 GB)?

Yes. Nemotron-Cascade 2 30B-A3B needs about 20 GB at Q4_K_M, inside a 24 GB card, at an estimated 160 tokens/sec.

How do I run Nemotron-Cascade 2 30B-A3B locally?

Install Ollama and run `ollama run nemotron-cascade-2:30b`. That pulls the weights and starts a local OpenAI-compatible endpoint; after the download nothing leaves the machine.

What other sizes does Nemotron Cascade 2 come in?

Nemotron-Cascade 2 30B-A3B (20 GB), Nemotron Cascade 2 70B (43 GB). Every size shares the family's training and licence; the larger ones score higher and need proportionally more memory.