Nex-AGI397B (17B active)Q4_K_M 下约 240 GB 显存

Nex-N2 Pro — 显存、速度与本地部署

作者: Jakub Rusinowski · 最后更新:

397B total, 17B active, Apache 2.0 — 240 GB at Q4_K_M, so datacenter or nothing. Terminal-Bench 2.1 at 75.3, SWE-bench Verified at 80.8, GPQA Diamond at 90.7, GDPval at 1585. Accepts image input as well as text. The permissive licence is the notable part: a 397B model you may deploy commercially without asking anyone.

Nex-N2 Pro 在 Q4_K_M 下约需 240 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91
创意86
编程91

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
131.3 GB—放不下
Q3_K_M
3.41 bpw
170 GB—放不下
Q4_K_M
4.83 bpw
240.5 GB—放不下
Q5_K_M
5.67 bpw
282.2 GB—放不下
Q6_K
6.56 bpw
326.3 GB—放不下
Q8_0
8.50 bpw
422.6 GB—放不下
F16
16.00 bpw
794.8 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. Nex-N2 Pro 显存计算器 →

运行 Nex-N2 Pro

目录中能运行 Nex-N2 Pro 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 Nex-N2 Pro

安装 Ollama,然后运行:

ollama run nex-n2
Hugging Face 上的权重: nex-agi/Nex-N2-Pro ↗

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
397 Billion (17B active)
上下文窗口
262,144
架构
Mixture-of-Experts (Qwen3.5-397B-A17B post-train, multimodal)
提供商
Nex-AGI
许可证
Apache 2.0
规格量化
Q4_K_M
系统内存
256 GB
记录更新于
2026-09-08
许可证Apache-2.0允许商业使用

Commercial use permitted. No usage restrictions beyond attribution.

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agentic taskssoftware engineeringenterpriseresearch

我的 GPU 能运行 Nex-N2 Pro 吗?

Nex-N2 的其他尺寸

Nex-N2 Pro — 常见问题

How much VRAM does Nex-N2 Pro need?

About 240 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 Nex-N2 Pro run on an RTX 4090 (24 GB)?

No. Nex-N2 Pro needs about 240 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 Nex-N2 Pro locally?

Install Ollama and run `ollama run nex-n2`. That pulls the weights and starts a local OpenAI-compatible endpoint; after the download nothing leaves the machine.

What other sizes does Nex-N2 come in?

Nex-N2 mini (22 GB), Nex-N2 Pro (240 GB). Every size shares the family's training and licence; the larger ones score higher and need proportionally more memory.