Muse Glimmer-30B
Muse
Parameters
30.0B
Architecture
—
Released
—
License
—
About
No detailed description available for this model.
Benchmark Scores
| Benchmark | Score | Date | Standing |
|---|---|---|---|
|
Terminal-Bench 2.1 (Terminus-2)
coding_agent
|
45.28%
|
18.08.2026 |
#36/44
better than 16%
|
|
GPQA Diamond
stem_reasoning
|
79.58%
|
18.08.2026 |
#55/80
better than 30%
|
|
Humanity's Last Exam
stem_reasoning
|
37.69%
|
18.08.2026 |
#45/71
better than 36%
|
|
IFBench (prompt loose)
instruction_following
|
74.36%
|
18.08.2026 |
#8/14
better than 46%
|
|
OmniDocBench 1.5
document_understanding
|
82.72%
|
18.08.2026 |
#20/26
better than 20%
|
|
CharXiv (RQ)
document_understanding
|
50.59%
|
18.08.2026 |
#14/34
better than 58%
|
|
AndroidWorld
general_agent
|
28.19%
|
18.08.2026 |
#7/13
better than 50%
|
|
Claw-Eval Pass^3
coding_agent
|
51.59%
|
18.08.2026 |
#11/21
better than 40%
|
|
OSWorld-Verified
general_agent
|
61.59%
|
18.08.2026 |
#14/25
better than 42%
|
|
Claw-Eval Avg
coding_agent
|
65.24%
|
18.08.2026 |
#27/40
better than 28%
|
|
SWE-bench Pro
coding_agent
|
57.08%
|
18.08.2026 |
#44/77
better than 41%
|
What is a good Terminal-Bench 2.1 (Terminus-2) score? What is a good GPQA Diamond score? What is a good Humanity's Last Exam score?
Hardware Requirements
Estimated VRAM per quantization at 4,096 tokens context
| Quantization | Weights | KV Cache | Total VRAM | On your hardware |
|---|---|---|---|---|
| FP16 — 16-bit | 60.0 GB | 4.7 GB | 67.7 GB | Which GPUs can run it? Calculate on your GPU |
| INT8 — 8-bit | 30.0 GB | 2.3 GB | 33.8 GB | Which GPUs can run it? Calculate on your GPU |
| INT4 — 4-bit | 15.0 GB | 1.2 GB | 16.9 GB | Which GPUs can run it? Calculate on your GPU |
| FP4 — 4-bit | 15.0 GB | 1.2 GB | 16.9 GB | Which GPUs can run it? Calculate on your GPU |
Only datacenter GPUs — FP16 needs 67.7 GB. For consumer hardware, choose the INT4 build with a multi-GPU setup.
Consumer GPUs that fit this model with a single card (INT4):
No comments yet
Be the first to share your thoughts!