Upstage250B (15B active)Q4_K_M 下约 152 GB 显存

Solar Open 2 250B-A15B — 显存、速度与本地部署

作者: Jakub Rusinowski · 最后更新:

Upstage's 250B mixture-of-experts with 15B active per token, built for agents, document-heavy work and coding. Three of every four layers use linear attention, so only 12 of 48 layers keep a KV cache and the 1M-token context costs a quarter of the cache of an all-softmax model. Upstage recommends vLLM for serving.

Solar Open 2 250B-A15B 在 Q4_K_M 下约需 152 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。

量化显存速度(估算)适配
Q2_K
2.63 bpw
83.5 GB—放不下
Q3_K_M
3.41 bpw
107.9 GB—放不下
Q4_K_M
4.83 bpw
152.3 GB—放不下
Q5_K_M
5.67 bpw
178.6 GB—放不下
Q6_K
6.56 bpw
206.4 GB—放不下
Q8_0
8.50 bpw
267.1 GB—放不下
F16
16.00 bpw
501.8 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. Solar Open 2 250B-A15B 显存计算器 →

运行 Solar Open 2 250B-A15B

目录中能运行 Solar Open 2 250B-A15B 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 Solar Open 2 250B-A15B

安装 Ollama,然后运行:

ollama run solar-open
Hugging Face 上的权重: upstage/Solar-Open2-250B ↗

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
250 Billion (15B active)
上下文窗口
1,048,576
架构
Mixture-of-Experts (320 experts, 8 active), [softmax, linear x3] x 12 hybrid attention
提供商
Upstage
许可证
Upstage Solar License
Availability
Weights downloadable
Practical deployment
Workstation (large unified memory or pro GPU)
规格量化
Q4_K_M
系统内存
256 GB
记录更新于
2026-11-11
许可证Upstage Solar 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agentsdocumentscodinglong context

Solar Open 2 250B-A15B — 常见问题

How much VRAM does Solar Open 2 250B-A15B need?

About 152 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 Solar Open 2 250B-A15B run on an RTX 4090 (24 GB)?

No. Solar Open 2 250B-A15B needs about 152 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 Solar Open 2 250B-A15B locally?

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