Muse Spark 1.3

Meta's Muse Spark 1.3 is a frontier multimodal reasoning model for long-horizon coding and agentic workflows, with strong gains in computer use, browsing, professional tool use, codebase understanding, instruction following, and million-token retrieval. It accepts text, images, audio, video, and files, supports tool calling and structured output, and always reasons before answering.

  • Reasoning
  • Vision
  • Audio Input
  • Video Input
  • Native PDF input
  • Tool Calling
  • Structured Output

Added Sep 2, 2026

Pricing

Auto routing · per 1M tokens
Input
$1.25
Output
$4.25
Cache read
$0.15
Compare provider prices

Specifications

Context window
1M
Max output
N/A
Avg output (7d)
1.3K tokens
Longer than 79% of models

Benchmarks

Sourced from Artificial Analysis.

Intelligence Index

48.1

Better than 97% of models compared

Coding Index

75.8

Better than 91% of models compared

Agentic Index

55.5

Better than 96% of models compared

Agentic work

AutomationBench-AA

Workflow automation with guardrail penalties

57.9%

Better than 80% of models compared

AutomationBench-AA Tasks Completed

Fully completed workflows without guardrail violations

29.8%

Better than 47% of models compared

AA-Briefcase

Agentic knowledge work (Elo)

1583 Elo

Better than 93% of models compared

GDPval-AA v2

Economically valuable tasks (Elo)

1684 Elo

Better than 95% of models compared

Document reasoning

GDP.pdf

Professional PDF reasoning: all-pass rate

26.6%

Better than 91% of models compared

AA-LCR v1.1

Long context reasoning with updated grading

83.0%

Better than 95% of models compared

MLCR-AA

Medical long-context reasoning

43.3%

Better than 85% of models compared

Reasoning

HLE

Humanity's Last Exam

48.7%

Better than 95% of models compared

CritPt

Research-level physics reasoning

24.9%

Coding

Terminal-Bench v4.0

Practical coding and terminal tasks

33.3%

Better than 83% of models compared

SciCode

Python programming for scientific computing

58.8%

Better than 92% of models compared

Knowledge

AA-Omniscience Accuracy

Proportion of correctly answered questions

43.6%

AA-Omniscience Hallucination Rate

Rate of incorrect answers among non-correct responses

32.9%

Legacy benchmarks

GPQA Diamond (legacy)

Graduate-level scientific reasoning

93.5%

Better than 97% of models compared

AA-LCR (unversioned / legacy)

Long context reasoning evaluation

83.0%

Better than 95% of models compared

GDPval-AA (unversioned / legacy)

Economically valuable tasks

59.0%

Last updated Oct 15, 2026

Artificial Analysis

Providers

Auto routing is available for this model. Explicit provider selection is not available.

Loading provider options…