Meta发布最强LLM Muse Spark 1.3，称已赶上Anthropic和OpenAI

UPDATED 22:41 EDT / SEPTEMBER 02 2026 AI Meta says it has caught up with Anthropic and OpenAI with Muse Spark 1.3, its most powerful AI model yet by Mike Wheatley Meta Platforms Inc. says it has more or less caught up with the biggest artificial intelligence labs with the release of its most powerful large language model so far, Muse Spark 1.3. The company said in a blog post today that the new model can be accessed by developers willing to pay for it through its application programming interface. It’s also going to be rolled out to users of Meta’s social media platforms, Facebook and Instagram, as well as the Meta AI application, in the coming days. In an interview with Bloomberg, Meta Chief AI Officer Alexandr Wang said Muse Spark 1.3 represents the company’s biggest jump in model performance so far, putting it at the same level as the most recent models created by OpenAI Group PBC and Anthropic PBC. Pointing to advances in Muse Spark 1.3’s coding and agentic automation capabilities, Wang said it is now “competitive” with Anthropic’s Claude Fable 5.1 and “better than” OpenAI’s GPT-5.6 Sol model, especially in terms of its ability to generate code. It also “outperforms any of the current Chinese models out there,” Wang claimed. Such claims are always difficult to judge when it comes to AI models, because they tend to be better at some tasks than others, and it’s easy for companies to game the benchmark tests they publish. However, an independent analysis by Artificial Analysis showed that Muse Spark achieved a score of 62 on its Intelligence Index after evaluating its performance across a host of popular benchmarks, putting it behind only Fable 5.1 and Opus 5, and ahead of OpenAI’s models. Meta has released Muse Spark 1.3, their fourth Muse Spark model release in five months. Muse Spark 1.3 (max), which is in limited preview for Meta’s partners, scores 62 on the Artificial Analysis Intelligence Index, behind only Claude Fable 5.1 and Claude Opus 5. The variant… pic.twitter.com/D7eL8bGwAU — Artificial Analysis (@ArtificialAnlys) September 2, 2026 The evaluation suggests that Meta could finally start seeing some payback from its massive, multibillion-dollar investments in AI. The company has been spending hundreds of billions of dollars on AI infrastructure and development in an effort to catch up with Anthropic, OpenAI and its Chinese rivals. Last year, founder and Chief Executive Mark Zuckerberg revamped Meta’s AI strategy, paying over $14 billion to acquire a stake in Wang’s former company ScaleAI Inc. and hire him to run his new Superintelligence Labs unit. Since then, Wang has stepped up Meta’s game by enhancing the capabilities of its models and launching new versions as part of an aggressive release cadence. However, Meta’s spending has drawn a lot of scrutiny from investors, who are getting nervous about when it will see a return on investment. That’s why Meta has abandoned the open-source model it followed with its Llama models in favor of a proprietary strategy similar to Anthropic and OpenAI. In July, when it launched Muse Spark 1.1, the company said it would be charging developers to access the model for the first time, though it was noticeably cheaper than its rivals’ models. Zuckerberg said at the time that the plan is to make the Muse Spark family one of the most affordable LLMs on the market. According to Wang, developers will not have to pay any more to access Muse Spark 1.3 than they were paying to use Muse Spark 1.2, which was launched in August. Like its two predecessors, it’s available through the Meta Model API, which also provides developers with various tools for building AI applications. Wang told Bloomberg that Meta has seen rapid adoption of the Muse Spark LLM family, with some developers using “trillions of tokens per week.” He said they’ll be very happy with Muse Spark 1.3, because it’s more efficient that version 1.2, using around 25% fewer tokens to accomplish the same tasks. It can also support multiple workflows at once, instead of requiring separate sessions for each one, and it’s better at handling long and complex instructions, and retaining context across multiple tasks. It also has more awareness of its own limitations, Wang said, and it is much safer. In every case when it’s about to take an action that’s irreversible, it will ask for confirmation before it goes and does it. Though Meta is now charging for access to its most powerful models, Zuckerberg recently authored a blog post stressing the need to make AI development more open and accessible. But the company has not yet decided if it’s going to release Muse Spark 1.3’s weights. Open-weights models, popularized by Chinese AI model makers, also release the “weights,” or the internal blueprint that shows how the model responds to user’s prompts. When companies release a model alongside its weights, it means that developers can download it and run it anywhere, and modify it in any way they see fit, without paying for it. Meta has previously promised to release the weights for Muse Spark 1.2, but has not yet done so. Meta may get around to doing this by the time it releases its highly anticipated model, called Watermelon, which is said to be larger and much more powerful than the Muse Spark models. The company has been working on Watermelon for some time, but Wang declined to say when it might be released. For now, it remains a work in progress, but Wang believes it will be worth the wait. “We believe that Watermelon will be extremely competitive,” he added. Image: Meta Platforms A message from John Furrier, co-founder of SiliconANGLE: Support our mission to keep content open and free by engaging with theCUBE community. Join theCUBE’s Alumni Trust Network, where technology leaders connect, share intelligence and create opportunities. 15M+ viewers of theCUBE videos, powering conversations across AI, cloud, cybersecurity and more 11.4k+ theCUBE alumni — Connect with more than 11,400 tech and business leaders shaping the future through a unique trusted-based network Are you an AWS customer?  Support SiliconANGLE financially by buying your AWS services from our Marketplace portal page and links: https://siliconangle.com/aws-marketplace/ About SiliconANGLE Media SiliconANGLE Media is a recognized leader in digital media innovation, uniting breakthrough technology, strategic insights and real-time audience engagement. As the parent company of SiliconANGLE, theCUBE Network, theCUBE Research, CUBE365, theCUBE AI and theCUBE SuperStudios — with flagship locations in Silicon Valley and the New York Stock Exchange — SiliconANGLE Media operates at the intersection of media, technology and AI. Founded by tech visionaries John Furrier and Dave Vellante, SiliconANGLE Media has built a dynamic ecosystem of industry-leading digital media brands that reach 15+ million elite tech professionals. Our new proprietary theCUBE AI Video Cloud is breaking ground in audience interaction, leveraging theCUBEai.com neural network to help technology companies make data-driven decisions and stay at the forefront of industry conversations. LATEST STORIES Former Uber CEO Travis Kalanick is returning to the robotaxi race OpenAI to set misalignment disclosure rules after agents took over a wiki What to expect during DigiCert's World Quantum Readiness Day event: Join theCUBE Sept. 17 What to expect at Workiva's Amplify event: Join theCUBE Sept. 15–16 Anthropic uses Claude to formalize proof of Fermat’s Last Theorem Equinix turns the network into the control plane for enterprise AI inference LATEST STORIES Former Uber CEO Travis Kalanick is returning to the robotaxi race EMERGING TECH - BY MIKE WHEATLEY . 9 HOURS AGO OpenAI to set misalignment disclosure rules after agents took over a wiki AI - BY DUNCAN RILEY . 10 HOURS AGO What to expect during DigiCert's World Quantum Readiness Day event: Join theCUBE Sept. 17 SECURITY - BY MARK ALBERTSON . 1 DAY AGO What to expect at Workiva's Amplify event: Join theCUBE Sept. 15–16 CLOUD - BY DEVONY HOF . 2 DAYS AGO Anthropic uses Claude to formalize proof of Fermat’s Last Theorem AI - BY MARIA DEUTSCHER . 2 DAYS AGO Equinix turns the network into the control plane for enterprise AI inference INFRA - BY ZEUS KERRAVALA . 2 DAYS AGO