PyyanThe AI Index

Pyyan / Compare / Beam 1.0 vs Muse Spark Open

Beam 1.0 vs Muse Spark Open

2 of 5

Open-Weight Models · verified 26 Oct 2026

RBeam 1.0Reflectioncurrent
Muse Spark OpenMetacurrent
3 slots left
SpecificationBeam 1.0Muse Spark Open
Summary501B under Apache 2.0, with nothing asked in return.The largest American open weights, under a licence with a revenue trigger.
Context1M1M
Input ($/Mtok)$0.25Weights only
Output ($/Mtok)$1Weights only
LicenceApache 2.0Muse Model License
Max output131KNot published
Parameters501B total, 23B active816B total, 42B active
Released16 Oct 202615 Oct 2026
Model IDreflection/beam-1.0-501b-a23bmeta-models/Muse-Spark-Open-816B-A42B
CategoryOpen-Weight ModelsOpen-Weight Models
OfficialReflection ↗Meta ↗

Highlighted rows are where these differ.

Beam 1.0

  • Apache 2.0, with no attribution clause and no revenue trigger, unlike Meta's Muse licence
  • 52 layers, 256 experts, a 1M token context and a knowledge cutoff of 30 June 2026
  • Reflection claims comparable answers to leading open models with roughly three times fewer reasoning tokens
  • Artificial Analysis scores it 30 on the intelligence index, 19th of 120 models

Best for open weights with no licence conditions attached.

Full spec sheet →

Muse Spark Open

  • Text, image, video and audio in, text out, with a 1M token context
  • Muse Model License, not open source: attribution required, and a separate agreement above $20m of model-service revenue
  • About 1.5TB in BF16, or two Mac Studios at 4-bit

Best for multimodal work you host yourself.

Full spec sheet →