Muse Glimmer 30B — 显存、速度与本地部署
作者： Jakub Rusinowski · 最后更新：
Text and image in, 128K context, Apache-2.0. Meta targets a 24 GB card with the 4-bit build; the Ollama tag is an 18 GB quantised download. Grouped-query attention keeps the KV cache small for its size.
Muse Glimmer 30B 在 Q4_K_M 下约需 19 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)。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 | 显存 | 速度（估算） | 适配 |
|---|---|---|---|---|
| Q2_K 2.63 bpw | 10.5 GB | ~61 tok/s | 可运行 | |
| Q3_K_M 3.41 bpw | 13.4 GB | ~50 tok/s | 可运行 | |
| Q4_K_M 4.83 bpw | 18.7 GB | ~37 tok/s | 可运行 | |
| Q5_K_M 5.67 bpw | 21.8 GB | ~32 tok/s | 勉强 | |
| Q6_K 6.56 bpw | 25.1 GB | ~4 tok/s | 需卸载 | |
| Q8_0 8.50 bpw | 32.3 GB | ~3 tok/s | 需卸载 | |
| F16 16.00 bpw | 60 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. Muse Glimmer 30B 显存计算器 →
运行 Muse Glimmer 30B
目录中能运行 Muse Glimmer 30B 的最便宜 GPU 是 AMD Radeon RX 7900 XTX (24 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 Muse Glimmer 30B
安装 Ollama，然后运行：
ollama run muse-glimmer:30b规格
Corroborated — Two or more independent sources agree on these figures, but the model card itself was not retrieved. Treat the numbers as good rather than confirmed.
- 参数量
- 29.6 Billion
- 上下文窗口
- 131,072
- 架构
- Dense + perception encoder
- 提供商
- Meta (Superintelligence Labs)
- 许可证
- Apache 2.0
- Practical deployment
- Single consumer GPU
- 规格量化
- Q4_K_M
- 系统内存
- 32 GB
- 记录更新于
- 2026-09-19
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。
Independent evaluation — not vendor-claimed and not measured by us
| 基准 | 分数 | 来源 |
|---|---|---|
| AA Intelligence Index v4.3.2 (High) | 17 | independent · Artificial Analysis · snapshot 2026-10-13 |
Muse Glimmer 30B — 常见问题
How much VRAM does Muse Glimmer 30B need?
About 19 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 Muse Glimmer 30B run on an RTX 4090 (24 GB)?
Yes. Muse Glimmer 30B needs about 19 GB at Q4_K_M, inside a 24 GB card, at an estimated 37 tokens/sec.
How do I run Muse Glimmer 30B locally?
Install Ollama and run `ollama run muse-glimmer:30b`. That pulls the weights and starts a local OpenAI-compatible endpoint; after the download nothing leaves the machine.