Model benchmarks

Muse-Glimmer-30B-KQuant-17GB-Q4_K_M local LLM performance

As of November 2026, Muse-Glimmer-30B-KQuant-17GB-Q4_K_M runs at up to 66.0 tok/s for local inference (best of 2 community benchmark runs across 0 GPUs).

llama.cppQ4_K_M
ShareRedditX

Model size

30B

Peak speed

66.0 tok/s

Average speed

65.9 tok/s

Avg PP

902.6 tok/s

Min memory

25.4 GB

Max context

131,072 tokens

Avg output / run

14,967 tokens

Avg runtime / run

3m 53s

Avg quality

83.7

Benchmark runs

2

GPUs tested

0

Quality by task

Average LLM-judged quality (0–100) with the run-to-run spread shown as a P5–P95 band, overall and for each benchmark task, across all 2 runs. The low and high columns show how much the judge’s score varies between runs, and need at least two runs to display.

TaskP5 (low)AvgP95 (high)
Overall82.283.785.2
Agent Workflow90.090.691.2
Code Generation71.874.076.2
Role Play & Narrative87.289.992.5
Research & Analysis80.080.480.9

Performance by hardware and tool

Every hardware/tool/quantization combination Muse-Glimmer-30B-KQuant-17GB-Q4_K_M has been benchmarked on, ranked by peak token generation speed. Last updated November 2026.

HardwareToolQuantBest tok/sAvg tok/sMemoryContextQualityRuns
CPU onlyllama.cppQ4_K_M66.0 tok/s65.9 tok/s25.4 GB131,072 tokens83.72

Benchmark runs

All 2 Muse-Glimmer-30B-KQuant-17GB-Q4_K_M runs submitted so far — expand one for its hardware, quality breakdown and per-scenario detail.

Frequently asked questions

Is Muse-Glimmer-30B-KQuant-17GB-Q4_K_M good for coding?
In our benchmarks, Muse-Glimmer-30B-KQuant-17GB-Q4_K_M scores 74.0/100 for coding. It runs at about 65.9 tok/s, so if you want more speed, turbo is faster (~124.1 tok/s) and still scores well for coding (85.3/100).
Is Muse-Glimmer-30B-KQuant-17GB-Q4_K_M good for agentic (tool-using) tasks?
In our benchmarks, Muse-Glimmer-30B-KQuant-17GB-Q4_K_M scores 90.6/100 for agentic workflows. It runs at about 65.9 tok/s, so if you want more speed, Qwen3.8-27B-IQ5R-V4-DFlash2-TP2-i8KV-r21 is faster (~133.3 tok/s) and still scores well for agentic workflows (91.8/100).
How fast is Muse-Glimmer-30B-KQuant-17GB-Q4_K_M for local inference?
Across 2 community benchmark runs, Muse-Glimmer-30B-KQuant-17GB-Q4_K_M reaches up to 66.0 tok/s and averages 65.9 tok/s.
How much memory does Muse-Glimmer-30B-KQuant-17GB-Q4_K_M need?
The leanest observed configuration used about 25.4 GB of memory (quantizations tested: Q4_K_M).
Which tools have been used to run Muse-Glimmer-30B-KQuant-17GB-Q4_K_M?
Benchmarks were submitted using llama.cpp. Results are community-contributed and updated as new runs arrive.