Skip to content
Model Pareto

Muse Spark Open

MetaOpen weightshuggingface.co ↗
Compare with nearest rivalsCompare with…
Input
—
Output
—
Speed
~58tok/sest.
Context
1Mtokens

Open-weights MoE successor line to Muse Spark 1.3; scores at xhigh reasoning effort.

Rankings

Self-hostingEstimated

How we estimate →

What it would cost to run these open weights yourself on rented GPUs. No API sells this model, so this is its price on the chart.

Hardware
16× H200 141GB
FP8 weights
Throughput
~27,021 tok/s
many requests batched
Price
$0.30–2.50 /M tok
busy → light use
On your own machine
Too large
needs data-center GPUs
Assumptions (5)
  • FP8 weights (816B params, 42B active per token) + 40% KV-cache headroom ≈ 1142 GB
  • 16× H200 141GB at $3.59–$7.91/GPU-hour on-demand (2026-09-24)
  • ~27,021 output tok/s aggregate at batch 1143 (bandwidth-bound); MoE compute scales with active params
  • Blended 3:1 input:output; prefill ~151,394 tok/s
  • Low = 75% utilization at the low GPU price; high = 20% at the high price

Benchmark scores

~ italic, dashed = no published score yet, estimated from related benchmarks.

Coding

SWE-bench VerifiedEstimated
~#3 of 20 (estimated)~85.1%
  1. #1Claude Opus 596.0%
  2. #2Claude Sonnet 585.2%
  3. ~#3Muse Spark OpenEstimated85.1%
  4. #3Beam 1.0 501B A23B81.6%
No published score — estimated from related benchmarks
SWE-bench Pro (Public)Estimated
~#8 of 23 (estimated)~64.5%
  1. #1Claude Opus 5.589.9%
  2. #2Claude Sonnet 5.581.3%
  3. #7Hy4 preview65.7%
  4. ~#8Muse Spark OpenEstimated64.5%
  5. #8GPT-5.6 Terra63.4%
No published score — estimated from related benchmarks

Hard reasoning

AIME (latest)Estimated
~#3 of 33 (estimated)~97.2%
  1. #1GLM-5.3 NVFP499.5%
  2. #2Beam 1.0 501B A23B97.7%
  3. ~#3Muse Spark OpenEstimated97.2%
  4. #3Inkling97.1%
No published score — estimated from related benchmarks

Agents

OSWorld 2.0Estimated
~#3 of 5 (estimated)~53.8%
  1. #1GLM-5.3 Flash59.1%
  2. #2Kimi K358.3%
  3. ~#3Muse Spark OpenEstimated53.8%
  4. #3GPT-5.6 Terra50.2%
No published score — estimated from related benchmarks
BrowseCompEstimated
~#14 of 22 (estimated)~83.2%
  1. #1Atria Dawn Preview92.5%
  2. #2GPT-6 Astra91.5%
  3. #13Nex-N2.5-Mini83.4%
  4. ~#14Muse Spark OpenEstimated83.2%
  5. #14Nemotron 3.5 Super VL78.0%
No published score — estimated from related benchmarks
GDPval-AA (Elo)Estimated
~#20 of 91 (estimated)~1514 Elo
  1. #1Claude Opus 5.51866
  2. #2Muse Wonder Full1857
  3. #19Kimi K31533
  4. ~#20Muse Spark OpenEstimated1514
  5. #20GPT-6 Sol1487
No published score — estimated from related benchmarks
AutomationBench-AAEstimated
~#28 of 84 (estimated)~40.5%
  1. #1Muse Wonder Full71.3%
  2. #2Claude Opus 5.569.5%
  3. #27GPT-5.547.3%
  4. ~#28Muse Spark OpenEstimated40.5%
  5. #28K2 Horizon 375B A23B37.2%
No published score — estimated from related benchmarks
τ-Bench Banking (AA)Estimated
~#22 of 91 (estimated)~31.4%
  1. #1Muse Spark 1.350.5%
  2. #2GLM-5.350.3%
  3. #21K2 Horizon MoVA 36B A4B32.0%
  4. ~#22Muse Spark OpenEstimated31.4%
  5. #22Inkling29.1%
No published score — estimated from related benchmarks