model comparison

vs

Beam vs Muse Spark 1.3 Contributor

Reflection against Meta, compared on listed price, real workload cost, context and capabilities.

the short answer

  • Muse Spark 1.3 Contributor lists lower input and output prices.
  • Muse Spark 1.3 Contributor supports vision; Beam does not.
  • Muse Spark 1.3 Contributor reads cached input for less ($0.002/M against Beam's $0.015/M), which decides the bill for agents that resend long context.

side-by-side specs

specBeamMuse Spark 1.3 Contributor
providerReflectionMeta
input price$0.125/M$0.1/M
output price$0.4/M$0.2/M
cached input$0.015/M$0.002/M
blended price (3:1)$0.194/M$0.125/M
context window1M tokens1M tokens
max output128k tokens—
inputtexttext, images, video, audio, pdf
structured outputyesyes
weightsopen (Apache 2.0)closed
releasedOctober 2026September 2026

same work, both models

workloadBeamMuse Spark 1.3 Contributor
one chat reply3,000 tokens in, 500 outBeam costs 1.4× as much as Muse Spark 1.3 Contributor$0.00058$0.0004
summarise a 100-page report50,000 tokens in, 1,500 out, one requestBeam costs 1.3× as much as Muse Spark 1.3 Contributor$0.0069$0.0053
1,000 classification calls500 tokens in, 20 out, per call$0.07$0.05
coding-agent session2M tokens in across ~60 turns, 60,000 out, no cache$0.27$0.21
same session, cached85% of input read from the prompt cache, the rest written to itBeam costs 1.9× as much as Muse Spark 1.3 Contributor$0.09$0.05

at listed rates. token counts are typical sizes, not measurements.

Muse Spark 1.3 Contributor pricing notes

  • Contributor rate: Meta may use prompts, files and responses to train future models; standard muse-spark-1.3 does not train on them.

when to pick which

Beam

the first open-weights model from Reflection, a 501B-total, 23B-active mixture-of-experts built for agentic tool use, coding and general reasoning. it reads text only and its weights ship under Apache 2.0.

pick it if

  • you want a long-window text model priced with the flash tier
  • you want Apache 2.0 weights, with BF16, FP8 and NVFP4 builds to self-host
  • you run tool-calling agents on text and repeat long prompts, so cached input matters

skip it if

  • you need image, audio or video input; Beam reads text only
  • long autonomous coding is the job; its reported DeepSWE v1.1 score sits below GLM-5.3 Flash and DeepSeek V4.1 Flash

Muse Spark 1.3 Contributor

Contributor mode of Meta's Muse Spark 1.3: the same model at $0.10 / $0.20 per 1M tokens instead of $1.25 / $4.25, in exchange for letting Meta train future models on your prompts and responses.

pick it if

  • you want Muse Spark 1.3 output at a fraction of the standard price
  • your prompts are non-sensitive, so Meta training on them is acceptable
  • you re-send long contexts; cached input bills at $0.002 per 1M

skip it if

  • prompts carry client code, secrets or personal data Meta must not train on
  • you need max reasoning effort or over 150 requests per minute

faq

which is cheaper, Beam or Muse Spark 1.3 Contributor?

Muse Spark 1.3 Contributor lists lower input and output prices. on a coding-agent session (2M tokens in, 60,000 out) Beam costs $0.27 and Muse Spark 1.3 Contributor $0.21.

which one accepts images?

Muse Spark 1.3 Contributor supports vision; Beam does not.

can i switch between Beam and Muse Spark 1.3 Contributor mid-task?

yes. in an octomind session, /model octohub:beam and /model octohub:muse-spark-1.3-contributor switch models without losing the conversation, so you can draft on one and review on the other.

try either model

switch between models inside the same octomind session.

open the panel all models

prices from the catalog of 2026-11-09