- Muse Spark 1.1 is a multimodal reasoning model built for agentic tasks, with major gains in tool and computer use, coding, and multimodal understanding.
- Meta unveiled a major update to its Muse Spark artificial intelligence model, offering pricing details that include $1.25 per million input tokens and $4.25 per million output tokens.
- The Muse Spark 1.1 model can actively manage its context window of 1 million tokens, allowing it to handle complex workflows effectively.
- Meta characterized Muse Spark 1.1 as its strongest model for agentic and coding work yet, with significant improvements in coding performance on real-world tasks.
- The public preview of Muse Spark 1.1 is now available to developers in the United States via Meta's Model API, allowing them to test prompts and build applications.
- Meta first introduced Muse Spark in April as a text-and-reasoning model, and it is now expanding access from a limited preview to a public preview.
Meta has unveiled Muse Spark 1.1, a significant upgrade to its AI model designed for coding and agentic tasks. This multimodal reasoning model boasts a one-million-token context window and enhanced capabilities in tool use and coding performance, making it a strong competitor against models from Anthropic and OpenAI.148
According to Meta, Muse Spark 1.1 delivers exceptional performance in personal agentic tasks, allowing it to plan and orchestrate actions across various applications. It can zero-shot generalize to new tools and custom skills, enhancing its versatility in real-world applications.
The model's coding capabilities have seen substantial improvements, particularly in handling large, complex codebases. On the internal coding evaluation, Muse Spark 1.1 significantly outperformed its predecessor and is competitive with leading alternatives. Alexandr Wang, Meta's AI chief, noted that the model is designed to work well with popular developer tools, aiming to maximize adoption.
Meta's pricing strategy for Muse Spark 1.1 is also noteworthy, with initial credits of $20 for new API accounts, followed by a pay-as-you-go model at $1.25 per million input tokens and $4.25 per million output tokens. This pricing is characterized as “very aggressive and attractive” compared to competitors.
The release of Muse Spark 1.1 marks a significant push by Meta into the enterprise AI market, expanding access to developers and enhancing its capabilities in autonomous workflows and coding tasks.
“Muse Spark 1.1 features significant improvements in coding performance, outperforming previous versions and competing well against leading models from Anthropic and OpenAI. The model is now available to US developers via Meta's Model API, starting with $20 in free credits before transitioning to a pay-as-you-go pricing model.”
CNBC
Seeking Alpha
