llama.cpp-local · reasoning
LLAMA.CPP-LOCAL

muse-glimmer-30b-q4_k_xl-local

Meta Muse Glimmer 30B, UD-Q4_K_XL on Owlforge RTX 3090 Ti at a 350 W power limit

⚖ COMPARE MODELS
★VALS INDEX
85.1% +/- 16.1
Accuracy
⏱VALS INDEX
34.2s
Latency (median)
$VALS INDEX
—
Cost / Test (1P-est)
SCORE STATUSscored
TEST SETS12
RECORDED TRIALS120
CUSTOM PASS80%
SEALED HRB v0.2 · LOCAL TRACK

45/48 correct · 93.75% pass@1

This is a separate 48-task HRB result, not a custom-suite score and not routing evidence.

Median wall
2.903 s
Cost
UNAVAILABLE
Receipt
3d3b6a3e845796bf…
HRB track and limitations →

Industry benchmarks

Accuracyattempts · responded
GPQA_DIAMOND GPQA Diamond
68.0%
25 attempts · 25 responded
IFEVAL IFEval
87.3%
25 attempts · 25 responded
HUMANEVAL_PLUS HumanEval+
100.0%
25 attempts · 25 responded
Industry benchmark Bars = accuracy %, black cap = that score

All trials on this page were run through the harness. Industry suites use deterministic family-specific reference graders (no LLM judge). The custom suite is graded by consensus of minimax-m3 + claude-fable-5. See /industry/ for the full benchmark suite.