Developers building computer-use agents with long context can use it, though Spark 1.3 has since replaced it at the same price.

Kenning's take

What it does

Muse Spark 1.1 is a multimodal reasoning model for agent tasks that can manage a context window of 1 million tokens, compacting earlier work while keeping the important steps. Meta says it is better at computer use across several applications and at work on large codebases. It launched in Thinking mode in the Meta AI app and on meta.ai, and through the Meta Model API in public preview.

• 1 million token context window with active compaction
• Computer use across multiple applications
• Better work on large codebases
• Image, video and audio understanding

Who it suits. Developers building agents on Meta's models, and Meta AI users of Thinking mode.

Key facts

Category

AI models

Part of

Meta AI

Made by

Meta (Meta Superintelligence Labs)

Released

9 July 2026

How to get it

Thinking mode in the Meta AI app and on meta.ai, and through the Meta Model API as muse-spark-1.1

Works on

Web, iOS, Android, API

API price

$1.25 per million input tokens, $4.25 per million output, $0.15 cached input (Standard tier); Contributor tier $0.10 input and $0.20 output in exchange for allowing training on prompts

FAQ

Facts from the maker's own pages and the sources listed. Checked 2026-10-11.