How much VRAM does Muse Glimmer 30B need?
Muse Glimmer 30B needs 16.8 GB for its weights at 4-bit, 18.0 GB with an 8K context and 18.3 GB with 32K. At 8-bit, 30.7 GB.
- Parameters
- 29.8B
- Longest context
- 128K
- Released
- Aug 2026
- License
- Apache 2.0
- Strongest at
- Docs
Memory at each precision
| Precision | Weights | With 8K | With 32K | With 128K |
|---|---|---|---|---|
| 16-bit | 55.5 GB | 56.7 GB | 57.0 GB | 58.2 GB |
| 8-bit | 29.5 GB | 30.7 GB | 31.0 GB | 32.2 GB |
| 6-bit | 22.7 GB | 24.0 GB | 24.3 GB | 25.5 GB |
| 5-bit | 19.7 GB | 20.9 GB | 21.2 GB | 22.5 GB |
| 4-bit | 16.8 GB | 18.0 GB | 18.3 GB | 19.6 GB |
| 3-bit | 13.6 GB | 14.8 GB | 15.1 GB | 16.3 GB |
| 2-bit | 10.4 GB | 11.6 GB | 11.9 GB | 13.1 GB |
| 1.6-bit | 5.7 GB | 6.9 GB | 7.2 GB | 8.5 GB |
One conversation, 16-bit context cache, runtime overhead included.
What runs it
| Machine | Best fit | Context | Speed | First reply |
|---|---|---|---|---|
| RTX 3060 12 GB | Too big | |||
| RTX 4060 Ti 16 GB | 3-bit | 40K | 14 tok/s | 2.8 s |
| RTX 3090 | 5-bit | 128K | 31 tok/s | 1.7 s |
| RTX 4090 | 5-bit | 128K | 33 tok/s | 0.7 s |
| RTX 5090 | 6-bit | 128K | 50 tok/s | 0.6 s |
| 2× RTX 3090 | 8-bit | 128K | 21 tok/s | 1.0 s |
| RTX PRO 6000 Blackwell | 8-bit | 128K | 39 tok/s | 0.5 s |
| M4 Max 128 GB | 8-bit | 128K | 11 tok/s | 9.7 s |
| M5 Max 128 GB | 8-bit | 128K | 13 tok/s | 2.8 s |
| Ryzen AI Max+ 395 128 GB | 8-bit | 128K | 5.0 tok/s | 5.6 s |
| DGX Spark 128 GB | 8-bit | 128K | 6.3 tok/s | 2.0 s |
| M3 Ultra 256 GB | 8-bit | 128K | 16 tok/s | 5.4 s |
| M5 Ultra 512 GB | 8-bit | 128K | 23 tok/s | 1.7 s |
| H100 SXM | 8-bit | 128K | 69 tok/s | 0.4 s |
| DGX Station GB300 | 8-bit | 128K | 131 tok/s | 0.2 s |
| 8× H200 NVL | 8-bit | 128K | 95 tok/s | 0.1 s |
Best fit is the highest quality, up to 8-bit, that leaves 16K of context. First reply is the wait for a 2K prompt. Speeds are rough, plus or minus 25%.
How to run it
- Ollama
ollama run muse-glimmer:30b- llama.cpp
llama-server -hf unsloth/Muse-Glimmer-30B-GGUF- vLLM
vllm serve meta-models/Muse-Glimmer-30B- MLX (Mac)
mlx_lm.chat --model mlx-community/Muse-Glimmer-30B-4bit
Questions
How much VRAM does Muse Glimmer 30B need?
Muse Glimmer 30B needs 16.8 GB for its weights at 4-bit, 18.0 GB with an 8K context and 18.3 GB with 32K. At 8-bit, 30.7 GB.
Can Muse Glimmer 30B run on an RTX 4090?
Yes. At 5-bit it holds 128K of context and writes about 33 tokens a second.