On Thursday, Meta introduced a revised version of Muse Spark, an innovative multimodal AI model focused on agentic coding, aiming to rival offerings from competitors like OpenAI and Anthropic.
The first iteration, known as Spark 1.1, was initially revealed in April. This advanced model is capable of multi-step reasoning, managing intricate tasks, overseeing digital workflows, and implementing new functionalities within enterprise systems, according to the company.
Although Meta is slightly trailing its rivals, who have been releasing similar technologies for some time, its entry into the sector still poses a competitive challenge.
A critical element of competition in the AI field is service pricing, and Meta seems to be presenting an attractive offer. The company plans to charge $1.25 for every million input tokens and $4.25 for output tokens, which is competitive but slightly higher than Anthropic’s Claude Haiku 4.5 and OpenAI’s GPT-5.6 Luna.
Meta emphasizes that Spark can efficiently manage extensive agentic tasks, troubleshoot issues, and facilitate large-scale code transitions—functions that businesses are increasingly turning to AI solutions for.
“Muse Spark 1.1 excels in personal agentic tasks that require intricate planning and coordination across various external applications and services,” the company stated in a blog entry.
Over recent years, Meta has rolled out several foundational AI models, and the launch of Muse Spark was significant enough for CEO Mark Zuckerberg to make a post on X for the first time in three years. His previous post was in July 2023, coinciding with the platform’s rebranding from Twitter to X.
In his update, Zuckerberg referred to Spark as “a robust agentic and coding model at a very low price,” highlighting its strengths in agentic tasks, tool interaction, and computer operations.
He also hinted at “more to come soon,” suggesting that additional models will be released in the near future.
This week has been significant for AI developments—especially for Meta, which also revealed a new AI image generation model called Muse Image on Tuesday. Other notable announcements include a new version of Grok from SpaceXAI and OpenAI’s latest model, GPT-5.6, which was also launched on Thursday. This underscores the ongoing vitality of competition in the AI landscape, with companies needing to work hard to differentiate themselves.
