Meta Platforms Inc. has unveiled its most powerful large language model to date, Muse Spark 1.3, which the company claims is now on par with the latest models from Anthropic and OpenAI.
In a blog post, Meta detailed that Muse Spark 1.3 is now available for developers willing to pay for access through its application programming interface (API). The model will also be integrated into Meta’s social media platforms, Facebook and Instagram, as well as the Meta AI application, within the coming days.
Meta Chief AI Officer Alexandr Wang stated in an interview with Bloomberg that Muse Spark 1.3 represents the company’s most significant advancement in model performance. Wang noted that the new model is competitive with Anthropic’s Claude Fable 5.1 and better than OpenAI’s GPT-5.6 Sol, particularly in its ability to generate code. The model is also superior to any of the current Chinese models, according to Wang.
An independent analysis by Artificial Analysis showed that Muse Spark 1.3 scored 62 on the company’s Intelligence Index, ranking it behind only Claude Fable 5.1 and Claude Opus 5. This places Meta’s model in a strong competitive position within the AI landscape.
Muse Spark 1.3 marks the fourth release in the Muse Spark family within five months. The model’s enhancements include improved coding and agentic automation capabilities, making it more efficient in handling complex instructions and retaining context across multiple tasks. According to Wang, developers will not have to pay more to access Muse Spark 1.3 than they were paying for Muse Spark 1.2, which was launched in August.
Meta has been pouring significant resources into AI development, with multibillion dollar investments in infrastructure and research. Last year, founder and CEO Mark Zuckerberg revamped Meta’s AI strategy by acquiring a stake in Wang’s former company, ScaleAI Inc., and establishing Superintelligence Labs. Wang has since stepped up the company’s AI efforts, releasing new versions of Muse Spark with an aggressive release cadence.
While Meta has shifted from an open source strategy to a proprietary approach, similar to Anthropic and OpenAI, the company has not yet released the weights for Muse Spark 1.3. This means developers cannot download and modify the model freely. However, Meta has promised to release the weights for Muse Spark 1.2, and Wang believes that the upcoming Watermelon model, which is significantly larger and more powerful, will be worth the wait.
Meta continues to emphasize the need for open and accessible AI development. Zuckerberg recently wrote a blog post stressing the importance of making AI more open and accessible. However, the company remains hesitant to fully release the weights for Muse Spark 1.3, despite its competitive standing.
In the broader tech landscape, other companies are also making significant strides in AI. Market research startup askpolly has raised $3 million to transform social media chatter into verifiable insights, while Google has launched two Gemini 3.8 models with cutting edge reasoning capabilities. As AI security concerns continue to rise, companies like HiddenLayer and Huskeys are securing substantial funding to address these challenges.
Meta’s move to proprietary AI models and its aggressive development strategy underscore the ongoing competition in the AI sector. With Muse Spark 1.3, Meta aims to bridge the gap with industry leaders like Anthropic and OpenAI, while maintaining its commitment to making AI more accessible.
Thinking about building an AI product?
Get in Touch