Muse Spark Open 816B A42B — Benchmarks

Benchmark scores for Muse Spark Open 816B A42B aggregated from public leaderboards (6 boards), with how it ranks among open models and how it compares with its nearest neighbours. See hardware requirements for what you need to run it.

llmrun Composite Score →

128.7

90% range 125.4–132.0 · from 5 boards

CodingNot enough data yet
AgentsNot enough data yet
Math128.490% range 125–132
ScienceNot enough data yet
ReasoningNot enough data yet

Muse Spark Open 816B A42B reasoning benchmarks

Multi-step logic and problem-solving — can the model think through a hard problem rather than recall an answer.

GPQA Diamond
GPQA Diamond is a set of extremely hard, graduate-level science questions (physics, chemistry, biology) written by domain experts and filtered so that skilled non-experts with web access still fail. It measures genuine reasoning rather than memorization.
88.6%
88.6%
median 54.0%open leader 93.1%
Overall best: Gemini 4 Argon (high) 96.4%
89th percentile · #12 of 98 open · #58 of 300 all
#12 of 98
#58 of 300
Independent
Epoch AI · Oct 2026

Muse Spark Open 816B A42B coding benchmarks

Writing, editing, and fixing real code — measured by whether the resulting program actually runs and passes tests.

LMArena WebDev
Code Arena (formerly WebDev Arena) is Arena's head-to-head coding leaderboard: two models each build a real web app or UI from the same prompt, and people vote blind on which result is better. The rating shown is a relative Bradley-Terry score, not a percentage — compare it only to other models on this board.
1511.3
1511.3
median 1411.4open leader 1655.2
Overall best: Claude Opus 5.5 (Max) 1807.3
76th percentile · #13 of 52 open · #55 of 138 all
#13 of 52
#55 of 138
Arena
Arena · Oct 2026

Muse Spark Open 816B A42B math benchmarks

Symbolic and quantitative problem-solving, from competition math to multi-step calculation.

AIME 2024/2025
AIME (American Invitational Mathematics Examination) is a prestigious high-school competition of hard, integer-answer problems. It's a widely-cited yardstick for multi-step mathematical reasoning, here on the 2024–2025 papers.
98.1%
98.1%
median 52.8%open leader 98.6%
Overall best: Claude Fable 5 (high) 100.0%
98th percentile · #3 of 86 open · #29 of 283 all
#3 of 86
#29 of 283
Independent
Epoch AI · Oct 2026
FrontierMath
FrontierMath is a benchmark of exceptionally hard, original research-level mathematics problems created with professional mathematicians. Even the strongest models solve only a small fraction, making it a frontier measure of genuine mathematical ability.
69.8%
69.8%
median 55.8%open leader 72.2%
Overall best: GPT 6 Astra Max 93.7%
94th percentile · #2 of 17 open · #28 of 114 all
#2 of 17
#28 of 114
Independent
Epoch AI · Oct 2026
FrontierMath Tier 4
FrontierMath Tier 4 is the hardest tier of Epoch AI's research-level mathematics benchmark, holding problems that working mathematicians consider genuinely difficult research questions. Scores here stay low even for frontier models, making it the sharpest available read on mathematical ability.
31.7%
31.7%
median 25.6%open leader 39.0%
Overall best: GPT 6.1 Sol Max 100.0%
92nd percentile · #2 of 13 open · #38 of 76 all
#2 of 13
#38 of 76
Independent
Epoch AI · Oct 2026

Muse Spark Open 816B A42B knowledge benchmarks

Breadth and depth of factual knowledge across many subjects, paired with the reasoning to apply it.

SimpleQA
SimpleQA measures factual accuracy on short, fact-seeking questions with a single correct answer — directly probing how often a model is right versus confidently wrong (hallucination) on simple facts.
56.6%
56.6%
median 36.5%open leader
Overall best: Gemini 4 Argon (high) 76.0%
top · #1 of 17 open · #19 of 91 all
#1 of 17
#19 of 91
Independent
Epoch AI · Oct 2026

Score vs VRAM in the ≤640 GB tier

Muse Spark Open 816B A42B
Other models
Best Score for its VRAM
x: VRAM (GB, log scale) · y: llmrun Score

#5 of 90 that fit ≤640 GB

Each dot is a model at the best quant that fits ≤640 GB. The line joins the models with the best Score for their VRAM; a dot below it is beaten by a model that needs no more memory.

Full ≤640 GB ranking →

Muse Spark Open 816B A42B vs its neighbours

Same boards for the nearest open models per llmrun Score. The highest score in each row is marked.

Board	Muse Spark Open 816B A42Bthis model	GLM 5.3nearest by ScoreQ4_K_M · 454 GB	DeepSeek V4 Pro 0813nearest by ScoreQ4_K_M · 991 GB	DeepSeek V4.1 Flashnearest by ScoreQ4_K_M · 458 GB
llmrun Score	128.7	130.2	129.2	129.1
GPQA Diamond	88.6%	90.9%	91.7%	89.8%
LMArena WebDev	1511.3	1621.1	1582.0	1618.4
AIME 2024/2025	98.1%	91.1%	98.6%	98.3%
FrontierMath	69.8%	68.8%	64.6%	67.4%
FrontierMath Tier 4	31.7%	29.3%	26.8%	26.8%
SimpleQA	56.6%	41.0%	52.9%	42.9%
Compare head to head		Compare	Compare	Compare

Scores aggregated from public benchmark sources (Epoch AI: 5 boards, Arena: 1 board) — each linked from the benchmark pages. llmrun does not run these benchmarks. Percentiles and ranks are among open models in the llmrun catalogue; the median and leader marks refer to the same field.