Muse Glimmer 30B
Files and memory at 8K context
| Quant | File | Files | + KV + runtime | Smallest fit | Source |
|---|---|---|---|---|---|
| Q4_K_M meta-models/Muse-Glimmer-30B-GGUF | 15.61 GiB | 1 | 16.8 GiB | 20 GB GPU | HF files |
| Q5_K_M bartowski/Muse-Glimmer-30B-GGUF | 18.72 GiB | 1 | 19.9 GiB | 24 GB GPU | HF files |
| Q8_0 bartowski/Muse-Glimmer-30B-GGUF | 27.58 GiB | 1 | 28.8 GiB | 32 GB GPU | HF files |
| MLX 4-bit mlx-community/Muse-Glimmer-30B-4bit | 18.08 GiB | 4 | 19.3 GiB | 32 GB Mac | HF files |
| MLX 8-bit mlx-community/Muse-Glimmer-30B-8bit | 31.08 GiB | 153 | 32.3 GiB | 48 GB Mac | HF files |
No verified file: FP16. Sizes are exact file bytes (text weights; projectors listed separately). “+” = recurrent state or other terms unknown. Smallest fit adds 0.5 GiB spare; Macs use 75% of memory.
Quality
Artificial Analysis Intelligence Index
17.5
Variant: Muse Glimmer (High) · observed 2026-10-17 · source
Rank #38 of 75 researched 2026 models.
Arena text rating
1,389
95% CI 1379–1399 · 3,897 votes · published 2026-10-15 · dataset
Exact listed model name; hosted reasoning/quantization may differ from local checkpoint.
Scores describe the evaluated hosted variant, not each quantized file. AA and Arena are never averaged.
Reported speeds 24
Reported by the linked sources, not reproduced by LocalFit. All rows in the speed database
Community threads
- Training LoRAs on Muse-Glimmer for llama.cpp: 5 things that cost us a week (label format, reasoning_strength, peg-native 500s, convert_lora_to_gguf crash, recipe) Hugging Face · 2026-09-14
- Tutorial on how to run with llama.cpp Hugging Face · 2026-08-13
Qualitative signals only; titles are not verified claims.