Meta Launches Muse Spark 1.1 AI Coding Tool with Competitive Pricing

Meta has released Muse Spark 1.1, its latest AI coding tool, as it seeks to strengthen its position in the AI sector. CEO Mark Zuckerberg announced the launch in his first X post since July 2023, highlighting the significance of this release. The AI coding market already includes established tools such as Claude Code and OpenAI’s Codex. To attract developers, Meta is focusing on competitive pricing for Muse Spark 1.1.

Key Highlights

• Meta launches Muse Spark 1.1 to compete in the AI coding tool market
• Muse Spark 1.1 offers lower pricing than rivals like OpenAI and Anthropic
• Model outperforms some competitors in benchmark coding tasks according to Meta
• Developers receive $20 in free credits for API access to Muse Spark 1.1
• Meta plans an open-source variant and is developing a more advanced model

Key Features and Capabilities

Muse Spark 1.1 is designed to handle a range of enterprise coding tasks. These include fixing bugs, managing digital workflows, deploying new features, and conducting large code migrations. Meta developed Muse Spark 1.1 through its Superintelligence Labs (MSL), led by Alexandr Wang. Wang explained that strong coding abilities improve AI agents’ capacity to perform multiple tasks autonomously. The model was trained to work with popular developer tools to encourage adoption.

When announcing Muse Spark 1.1, Zuckerberg shared benchmark results comparing it to other AI models. According to Meta, Muse Spark 1.1 outperformed Anthropic’s Opus 4.8 and OpenAI’s GPT-5.5 in certain coding tasks. However, the benchmarks did not include Anthropic’s Mythos and Fable models or OpenAI’s GPT-5.6.

Pricing and Industry Response

Meta emphasizes Muse Spark 1.1’s low cost as a key advantage. Developers can access the tool via the Meta Model API. Each new API account starts with $20 (about Rs 1,900) in free credits. After that, Meta charges $1.25 (about Rs 120) per million input tokens and $4.25 (about Rs 405) per million output tokens.

In comparison, SpaceXAI’s Grok 4.5, released a day before Muse Spark 1.1, costs $2 (about Rs 190) per million input tokens and $6 (about Rs 570) per million output tokens. Anthropic’s Claude Opus 4.8 is priced at $5 (about Rs 480) per million input tokens and $25 (about Rs 2,390) per million output tokens. OpenAI’s GPT 5.6 Terra charges $2.5 (about Rs 240) per million input tokens and $15 (about Rs 1,430) per million output tokens.

Elon Musk responded to Zuckerberg’s announcement on X with the comment, “Jinx.” Musk’s SpaceXAI had just launched Grok 4.5, its first coding AI model.

Industry Developments and Meta’s Strategy

The release of Muse Spark 1.1 comes during a busy period for the AI industry. Meta also launched Muse Image, an image-generation model, through the Meta AI app, WhatsApp, and Instagram. OpenAI released its GPT-5.6 models and introduced ChatGPT Work, an AI agent for workplace tasks.

Muse Spark 1.1 marks a shift in Meta’s AI business model. While Meta previously focused on open-source models like Llama, it now charges developers for access to this proprietary tool. Alexandr Wang stated that Meta remains committed to open source and is developing an open-source version of Muse Spark, though no release date was given. Wang also mentioned that Meta is training a more advanced model, code-named Watermelon.