NOUMENOpen app
NOUMEN

Meta · Frontier

Muse Spark Open

Added Oct 15, 2026

Compare models

Choose up to 3 models to compare.1 of 3 chosen

Loading models…

Use in the app

Built with Muse Spark Open. Meta's open-weight 816B-A42B model for coding and agentic workflows. Text and image input, text output, tool calling and structured output. Always reasons at low, medium, high, or xhigh effort. Custom Muse Model License and Usage Policy apply.

  • Reasoning
  • Tools
  • Vision
  • Structured output
Input
Text, Images
Output
Text
Reasoning effort
low · medium · high · xhigh
Discovery tier
Frontier

Pricing

USD per 1M tokens
Input
$1.25
Output
$4.25
Cache read
$0.15

Reference prices as published by the model provider. API example

Specifications

Context window
1M
Max output
—
Average output (7 days)
1.3K tokens
Longer than 80% of 473 models

Side by side

Muse Spark Open compared

Specifications, prices, speed and benchmark scores next to each other. Benchmarks below show every compared model too.

Compare models

Choose up to 3 models to compare.1 of 3 chosen

Loading models…

Muse Spark OpenMetaThis modelGPT 5.5OpenAIRemove
ProviderMetaOpenAI
Discovery tierFrontierFrontier
AddedOct 15, 2026Apr 23, 2026
Limits
Context window1.05M1.05MBest
Max output—128K
Average output1.3K tokens490 tokens
Pricing, USD per 1M tokens
Input$1.25Best$5.00
Output$4.25Best$30.00
Cache read$0.15Best$0.50
Performance
Request success (24h)—100.0%
Benchmark indexes
Intelligence Index38.9Best38.4
Coding Index—74.9
Capabilities
ReasoningYesYes
ToolsYesYes
VisionYesYes
Structured outputYesYes
PDF inputNoYes
Reasoning effortlow, medium, high, xhighnone, low, medium, high, xhigh

Public evaluations

Benchmarks

Independent evaluation results, with how each score ranks among the models in the Noumen catalog.

  • Muse Spark Open
  • GPT 5.5

Intelligence Index

  • Muse Spark Open38.9
  • GPT 5.538.4

Muse Spark Open tested at xhigh reasoning effort

Muse Spark Open: Better than 83% of 326 scored models

3.8Catalog models by score55.1

Indexes

  • Coding Index

    Composite of the coding evaluations

    • Muse Spark Open: —
    • GPT 5.5: 74.9

Agentic work

  • AutomationBench-AA

    Workflow automation, penalized for guardrail violations

    • Muse Spark Open: 49.7%Top
    • GPT 5.5: 47.3%

  • AutomationBench-AA tasks completed

    Workflows finished end to end without a guardrail violation

    • Muse Spark Open: 20.9%Top
    • GPT 5.5: 19.6%

  • AA-Briefcasev1.1

    Agentic knowledge work, rated head to head

    • Muse Spark Open: 1,279 EloTop
    • GPT 5.5: 1,136 Elo

  • GDPval-AAv2.1

    Economically valuable occupational tasks, rated head to head

    • Muse Spark Open: 1,401 EloTop
    • GPT 5.5: 1,351 Elo

Document reasoning

  • GDP.pdf

    Professional reasoning over PDF documents, all-pass rate

    • Muse Spark Open: 18.6%
    • GPT 5.5: 21.2%Top

  • AA-LCRv1.1

    Reasoning across long documents

    • Muse Spark Open: 78.3%
    • GPT 5.5: 84.3%Top

  • MLCR-AA

    Long-context reasoning over medical records

    • Muse Spark Open: 7.8%
    • GPT 5.5: 28.9%Top

Reasoning and knowledge

  • Humanity’s Last Exam

    Expert-level questions across academic fields

    • Muse Spark Open: 43.9%
    • GPT 5.5: 45.8%Top

  • GPQA Diamond

    Graduate-level science questions

    • Muse Spark Open: 89.0%
    • GPT 5.5: 93.5%Top

  • IFBench

    Following precise output instructions

    • Muse Spark Open: —
    • GPT 5.5: 75.9%

Coding

  • Terminal-Benchv4.0

    Practical software engineering in a terminal

    • Muse Spark Open: 8.1%
    • GPT 5.5: 14.6%Top

  • SciCode

    Scientific computing in Python

    • Muse Spark Open: 58.4%Top
    • GPT 5.5: 55.8%

Legacy evaluations

  • Terminal-Bench Hard

    Earlier terminal task suite

    • Muse Spark Open: —
    • GPT 5.5: 60.6%

  • τ²-Bench Telecom

    Tool use in a customer-support simulation

    • Muse Spark Open: —
    • GPT 5.5: 93.9%

  • AA-LCR (unversioned)

    Long-context reasoning, earlier grading

    • Muse Spark Open: 78.3%
    • GPT 5.5: 84.3%Top

Scores from Artificial Analysis, updated Oct 26, 2026. Muse Spark Open was tested at xhigh reasoning effort; compare scores measured at the same effort.Artificial Analysis

Live routing data

Performance and reliability

Recent averages across requests routed for this model. Your own speed depends on prompt length, reasoning effort and load.

Prompt cache hits
0.0%
About $1.25 per 1M input tokens with caching

Cache hits are the share of input tokens recently served from a prompt cache across all requests. They are not a forecast of your own hit rate.

Developers

Use it through the API

Noumen’s API is OpenAI-compatible: any Chat Completions client works with a base URL and a key.

Read the quickstart
curl https://www.noumen.xyz/api/v1/chat/completions \
  -H "Authorization: Bearer $NOUMEN_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "meta/muse-spark-open",
    "reasoning_effort": "medium",
    "messages": [{ "role": "user", "content": "Explain this stack trace." }]
  }'
import os
from openai import OpenAI

client = OpenAI(base_url="https://www.noumen.xyz/api/v1", api_key=os.environ["NOUMEN_API_KEY"])
reply = client.chat.completions.create(
    model="meta/muse-spark-open",
    messages=[{"role": "user", "content": "Explain this stack trace."}],
)
print(reply.choices[0].message.content)

Meta

Other Meta models, starting with the same family.