Meta says it has caught up with Anthropic and OpenAI with Muse Spark 1.3, its most powerful AI model yet

Meta says Muse Spark 1.3 is its strongest AI model yet, and the company is pitching it as proof that it has finally closed the gap with Anthropic and OpenAI. The model is already available to paying developers through Meta’s API, and it’s also being pushed to Facebook, Instagram, and the Meta AI app in the coming days.

Chief AI officer Alexandr Wang told Bloomberg that the jump in performance is Meta’s biggest so far. He said Muse Spark 1.3 is competitive with Anthropic’s Claude Fable 5.1 and better than OpenAI’s GPT-5.6 Sol, especially for code generation. He also claimed it beats current Chinese models.

Those kinds of claims are always slippery, because AI models can look brilliant in one area and merely decent in another. But an independent evaluation by Artificial Analysis gave Muse Spark a score of 62 on its Intelligence Index, putting it behind only Fable 5.1 and Opus 5, and ahead of OpenAI’s models. That at least gives Meta something more solid than a victory lap.

The new model is also supposed to be cheaper to run than Muse Spark 1.2, using around 25% fewer tokens for the same tasks. Meta says it can handle multiple workflows at once, deal with longer instructions, and keep track of context across more than one task. Wang also said it is safer, with confirmation required before irreversible actions.

Meta’s AI push has been expensive enough to make investors nervous, and the company has leaned hard into a proprietary strategy after its earlier open-source approach with Llama. It started charging developers for Muse Spark 1.1 in July, kept the price steady for Muse Spark 1.3, and still hasn’t decided whether to release the model’s weights. Watermelon, the bigger model Meta has been teasing, remains unfinished.