PyyanThe AI Index

Pyyan / Language Models / Muse Spark 1.3

Muse Spark 1.3

A quarter fewer tokens than 1.2 for the same work, which is the figure on the bill.

currentMetaVerified 2026-09-05

88.8

Terminal-Bench 2.1

Specifications

Context1Msource · as of 2026-09-05 · medium confidence
Input ($/Mtok)$1.25source · as of 2026-09-05 · medium confidence
Output ($/Mtok)$4.25source · as of 2026-09-05 · medium confidence
LicenceProprietarysource · as of 2026-09-05
Max outputNot publishedsource · as of 2026-09-05
ParametersNot publishedsource · as of 2026-09-05
Released2 Sep 2026source · as of 2026-09-05
Model IDmuse-spark-1.3source · as of 2026-09-05 · medium confidence

Official links

Muse Spark 1.3 model page

What matters

  • About 20% fewer tool calls and 25% fewer tokens than 1.2 to reach the same result
  • 75.4 on DeepSWE v1.1 and 98.5 on MRCR long context between 256K and 512K
  • A Contributor tier costs $0.10 in and $0.20 out for the same model, in exchange for letting Meta train on your prompts

Best for long running agentic coding.

Compare with

All 5 side by side →

Muse Spark 1.3 vs GPT-6 Astra →Muse Spark 1.3 vs Claude Fable 5.1 →Muse Spark 1.3 vs Claude Fable 5 →Muse Spark 1.3 vs Claude Opus 5 →Muse Spark 1.3 vs GPT-5.6 Sol →

Also in Language Models

GPT-6 Astra · OpenAIClaude Fable 5.1 · AnthropicClaude Fable 5 · AnthropicClaude Opus 5 · AnthropicGPT-5.6 Sol · OpenAI