Zuck's Muse to Spark Joy with Open Weights Release 'Soon'
The model uses 25% fewer tokens and adds stronger safety controls as Meta pushes to keep pace with rival AI systems.
- Meta released its most powerful AI model, Muse Spark 1.3, with plans to roll out the update soon to users of Instagram, Facebook, and Meta AI.
- Featuring 25 per cent fewer tokens, the new model improves efficiency while managing multiple workflows and enhances safety by seeking clarification before taking irreversible actions.
- Wang described Muse Spark 1.3 as "competitive" with Anthropic's Claude Fable 5.1 and "better than" OpenAI's GPT-5 Sol in coding, though OpenAI will soon release Astra.
- Meta CEO Mark Zuckerberg recently emphasized making AI development more open and accessible, though the company must balance these goals against massive infrastructure investments.
- Development remains on track for Meta's highly anticipated larger model dubbed Watermelon, which the AI chief believes will be extremely "competitive" upon release.
18 Articles
18 Articles
Meta's new AI model edges closer to OpenAI and Anthropic, its AI Chief says
Meta has released Muse Spark 1.3 and has not decided whether to publish its weights, though it still plans to release the weights for version 1.2. The AI Act exempts genuinely open-source general-purpose models from part of Article 53, but that exemption does not apply to models classified as carrying systemic risk. Meta has released […] This story continues at The Next Web
This version produces cleaner and more efficient output in programming tests than the previous generation, with a 25% reduction in token consumption. The post Meta Unveils Muse Spark 1.3 for Advanced Agent-Driven and Coding Tasks appeared first on Digiato.
The new model integrates directly into the Muse Code terminal tool and the Meta Model API, maintaining exactly the same cost structure as the previous version, making it highly affordable for professionals.
Meta launched Muse Spark 1.3 on the same day as Google's Gemini 3.8 Flash. Its sales argument: take down the same coding work by consuming less, thus costing developers less.
Coverage Details
Bias Distribution
- 50% of the sources are Center
Factuality
To view factuality data please Upgrade to Premium
