Mark Zuckerberg Posts After Three Years of Disappearance, All for This: Meta's Most Powerful Agent Model Enters the Field of Programming

Three months after launching its first AI model under the leadership of AI chief Alexandr Wang, Meta has rolled out another major upgrade, attempting to compete with OpenAI and Anthropic in key segments of the AI market.

Yesterday, Meta officially released the new Muse Spark, a multimodal AI model built for agentic coding, designed to rival comparable offerings from companies like OpenAI and Anthropic. In an interview, Wang stated that Muse Spark 1.1 is "the most capable model available today for agentic tasks and coding."

Over the past several years, Meta has sequentially launched a series of foundational AI models. The release of Muse Spark is evidently significant enough that it prompted CEO Mark Zuckerberg to make his first post on the X platform in three years. Zuckerberg's previous post dates back to July 2023, shortly after the platform rebranded from Twitter to X.

In the post, he described Spark as "an extremely low-cost yet powerful agent and coding model," noting that it "excels at agent performance, tool calling, and computer operation." Zuckerberg also teased that "there is much more to come soon," signaling Meta's plan to continue rolling out additional models.

Remarkable Performance Leap in Three Months, "Outperforming Competitors on Agentic Tasks" 

Meta's core selling point to users is Spark's ability to handle large-scale agentic tasks, including fixing code vulnerabilities and assisting with large-scale code migrations — capabilities that an increasing number of enterprises are seeking to implement via AI solutions. As a multimodal reasoning model purpose-built for agentic tasks, Muse Spark 1.1 delivers substantial improvements in tool and computer usage, coding, and multimodal understanding.

In its official blog, Meta wrote: "Muse Spark 1.1 demonstrates exceptional performance in personal agent tasks, particularly suited for scenarios requiring planning and coordination across multiple external applications and services." In multi-app computer usage workflows, Muse Spark 1.1 performs outstandingly by retaining context across long sessions and intelligently selecting between scripts, direct interface interactions, and batch operations at each step. It can navigate unfamiliar interfaces with minimal human intervention. Wang revealed that in certain tasks requiring interaction with various third-party programming products and services, Muse Spark 1.1 outperforms competing models.

According to Wang, Meta Superintelligence Labs (MSL), which he leads, trained Muse Spark 1.1 specifically on programming-related tasks, as this ultimately enhances the overall capabilities of AI agents, enabling them to autonomously execute multiple tasks like "a cohort of human interns." "You have to build coding proficiency as an integral part of overall agent capability," Wang stated.

Wang also shared that he has been "dog-fooding" the latest version of Muse Spark internally, and is excited about its potential as a personal health enhancement tool. For example, the model can help users search the web, read academic papers, and access personal health-related data. Speaking about his experiments using AI to assist with health management, Wang said, "This is exactly the kind of use case that I believe truly demonstrates the demand for agentic systems."

It is understood that Muse Spark originally had the internal codename Avocado, with its first generation launched in April this year. Meta stated that the initial Spark 1.1 version announced in April features multi-step reasoning capabilities, can handle complex workflows, manage digital workstreams, and deploy new functionalities in enterprise systems. Compared to the original model, Muse Spark 1.1 achieves a significant performance leap in implementing complex features, end-to-end development tasks, and codebase searching and comprehension.

Meta revealed that Muse Spark 1.1 is currently widely deployed across Meta's coding and research workflows, competing against leading models in Meta's internal coding benchmarks. Its researchers now leverage Muse Spark 1.1 in their workflows to automate model development and evaluation tasks. In perception and multimodal reasoning, Muse Spark 1.1 also delivers strong results, capable of examining visual and audio inputs, retaining details across lengthy workflows, and acting in real-world execution environments — it demonstrates particular strengths in visual-to-code generation, rich image/video captioning, and intelligent computer usage.

Entering the AI Coding Market with "Extremely Aggressive and Attractive" Pricing 

In this segment, Meta is currently slightly behind its competitors. Anthropic and OpenAI have offered similar models for some time. But this does not mean Meta's entry will not pose a threat.

A longstanding core competitive focus in the AI industry is the usage cost of models, and Meta appears to be targeting the market this time via a pricing advantage. According to foreign media reports, Meta will charge $1.25 per million input tokens and $4.25 per million output tokens for Spark. This price point is higher than OpenAI's entry-level model GPT-5 mini and Anthropic's low-cost model Claude Haiku 4.5, roughly on par (slightly higher) with GPT-5.6 Luna, but lower than Anthropic's more premium Claude Sonnet 4.6.

Wang stated that the updated pricing for Muse Spark is "extremely aggressive and attractive" compared to comparable offerings from labs like Anthropic and OpenAI. It is reported that every new API account receives a $20 free credit for model testing before moving to a pay-as-you-go billing structure. "Our goal is to deliver truly compelling pricing that can scale to meet large-scale usage demands."

Moreover, Wang noted that when Meta trained Muse Spark 1.1, it ensured the model "integrates seamlessly with all the mainstream toolchains developers currently use," which Meta believes is the best way to achieve maximum widespread adoption of the model.

"If Muse Spark 1.1 can genuinely compete with Claude and GPT in coding capabilities, Meta may finally have found a clearer commercialization path to convert its AI models into paid developer tools," said Shay Boloor, chief market strategist at Futurum Equities.

Planned to Replace Some Llama Models, Open-Source Version in Development 

Previously, the first generation of Muse Spark was only available to "select partners," who could access the technology exclusively via a "private API preview." Currently, U.S.-based developers can access Muse Spark through the Meta Model API public preview to test prompts, compare model outputs, and develop prototype integrations. Meta has now opened the public preview of the new model API through its developer portal, where users can register and review integration guides. A Meta spokesperson stated that some early partners already have API access, while new users "can join a waitlist and will be granted access incrementally."

The new model is already live in the "Thinking mode" of the Meta AI app and website. Additionally, Muse Spark is expected to replace some existing Llama models that currently power the chatbots on WhatsApp, Instagram, Facebook, as well as Meta's smart glasses product line. Meta indicated that it is still restricting API access within its own product ecosystem, and has not opened up to third-party platforms like the popular OpenRouter model marketplace. "This service will run on the compute infrastructure we have already built out," Wang said.

Notably, Meta's previous AI strategy primarily focused on making the Llama series of models available to the open-source community, but the company is now shifting toward selling access to its in-house developed AI models. However, Wang emphasized that Meta remains "committed to open source," and revealed that his MSL team is developing a "variant version of Muse Spark" that is planned for future open-source release, though he declined to disclose the specific launch timeline for that version.

Apart from Muse Spark, Meta recently released a new AI image generation model called Muse Image, which was previously codenamed "Mango." This model is designed to help Meta attract creators and advertisers to use its AI products.

According to Wang's latest disclosure, Meta is currently training an even more powerful AI model codenamed Watermelon, which has caught up with OpenAI's GPT-5.5 on key benchmarks, though no release date has been announced yet.

References:

https://techcrunch.com/2026/07/09/meta-enters-the-crowded-ai-coding-battle-with-muse-spark-1-1/

https://www.reuters.com/business/meta-debuts-muse-spark-11-with-preview-open-developers-2026-07-09/

This article is sourced from the WeChat public account "AI Frontline", compiled by Hua Wei, and republished with authorization from 36Kr.