{"id":1653092,"date":"2026-06-05T15:16:42","date_gmt":"2026-06-05T19:16:42","guid":{"rendered":"https:\/\/observer.com\/?p=1653092"},"modified":"2026-06-05T15:16:42","modified_gmt":"2026-06-05T19:16:42","slug":"alexandr-wang-defends-meta-muse-spark-model","status":"publish","type":"post","link":"https:\/\/observer.com\/2026\/06\/alexandr-wang-defends-meta-muse-spark-model\/","title":{"rendered":"Alexandr Wang Defends Meta’s Muse Spark as an \u2018Appetizer\u2019 in Bigger A.I. Push"},"content":{"rendered":"
\"Alexandr
Wang says Muse Spark is an early step, not Meta\u2019s final answer in the A.I. race. Courtesy Bloomberg<\/span><\/figcaption><\/figure>\n

In early April, Meta unveiled Muse Spark<\/a>, its latest A.I. model and the first major release under Chief A.I. Officer Alexandr Wang<\/a>. The model performs competitively on some benchmarks but still trails OpenAI\u2019s GPT-5.4 Pro and Google\u2019s Gemini 3.1 Pro, fueling questions about whether Meta has fallen behind. Wang believes that framing misses the point.<\/p>\n

“The new Muse Spark model that we released is not at the tier of the leading frontier models,” Wang said during an onstage interview at Bloomberg Tech Summit in San Francisco yesterday (June 4). “But we believe it’s a very exciting data point on the trajectory, and we expect the upcoming models we release to be quite competitive with the leading models in the world.”<\/span><\/p>\n

He described Muse Spark as an \u201cappetizer.\u201d Asked when the entr\u00e9e would arrive, Wang replied: \u201cWe\u2019re cooking it. We\u2019re seeing very exciting and promising results in the process of training it right now.\u201d<\/p>\n

Muse Spark marks a shift for Meta. It is the company\u2019s first proprietary model and is deployed only in Meta\u2019s products, rather than being released openly as prior systems were. The model is designed to handle text, images, video and audio, and to support more complex, multi-step tasks, including shopping features tied to Instagram and Facebook content.<\/p>\n

The release follows a difficult stretch for Meta\u2019s A.I. efforts. Llama 4, launched in April 2025, was widely criticized. Two months later, Mark Zuckerberg hired Wang to lead its newly formed Superintelligence Labs and reset its strategy.<\/p>\n

Wang said the group is focused on scaling: expanding data, computing power and research to drive improvements. And Muse Spark sits early in that process.<\/p>\n

The barrier to the frontier isn\u2019t money, Wang said. \u201cIt\u2019s about continuing to scale the data, the compute \u2026 as well as continuing to scale with research. All of the labs are dramatically scaling up their models, and we\u2019re on a much faster trajectory because we\u2019ve been doing all this work over the past year.\u201d<\/p>\n

Meta is backing that approach with heavy spending. The company expects capital expenditures of $125 billion to $145 billion in 2026, up from $72.2 billion in 2025, and is targeting more than 1.3 million GPUs and roughly one gigawatt of A.I. computing capacity.<\/p>\n

The shift to a closed model also reflects safety concerns. During development, Muse Spark triggered internal alerts, including around potential biological risks.<\/p>\n

\u201cWhen the company launches a model in a product, we have a lot of ways to mitigate some of these risks,\u201d he said. \u201cIt\u2019s much harder to do that when you open-source the model.\u201d<\/p>\n

Meta hasn\u2019t abandoned open-source A.I. entirely and continues to develop models it considers safe to release. Whether its Llama brand will continue remains undecided. \u201cWe have exciting debates about branding internally,\u201d Wang said, \u201cand nothing to share right now.\u201d<\/p>\n

Wang said Muse Spark\u2019s strengths are in multimodal capabilities, health-related applications and creative coding\u2014such as generating simple games or digital tools. Those areas underpin Meta\u2019s broader push into A.I. agents.<\/p>\n

The company is \u201cdoubling down\u201d on agents, aiming to build what Wang called \u201cthe best personal agents for everybody around the world.\u201d He said he uses such tools himself to manage his health and stay in touch with friends.<\/p>\n

The push comes alongside major internal changes. In May, Meta notified roughly 8,000 employees of layoffs and reassigned about 7,000 others to A.I.-focused roles as part of a broader reorganization.<\/p>\n

\u201cIt\u2019s incredibly difficult to say goodbye to teammates,\u201d Wang said. \u201cWe don\u2019t take any of it lightly.\u201d<\/p>\n","protected":false},"excerpt":{"rendered":"

Alexandr Wang defended Meta\u2019s Muse Spark and said the company\u2019s next models will be more competitive.<\/p>\n Read More<\/a>","protected":false},"author":129757851,"featured_media":1653158,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"_acf_changed":false,"apple_news_api_created_at":"","apple_news_api_id":"","apple_news_api_modified_at":"","apple_news_api_revision":"","apple_news_api_share_url":"","apple_news_cover_media_provider":"image","apple_news_coverimage":0,"apple_news_coverimage_caption":"","apple_news_cover_video_id":0,"apple_news_cover_video_url":"","apple_news_cover_embedwebvideo_url":"","apple_news_is_hidden":"","apple_news_is_paid":"","apple_news_is_preview":"","apple_news_is_sponsored":"","apple_news_maturity_rating":"","apple_news_metadata":"\"\"","apple_news_pullquote":"","apple_news_pullquote_position":"","apple_news_slug":"","apple_news_sections":[],"apple_news_suppress_video_url":false,"apple_news_use_image_component":false,"_jetpack_memberships_contains_paid_content":false,"footnotes":"","jetpack_publicize_message":"","jetpack_publicize_feature_enabled":true,"jetpack_social_post_already_shared":true,"jetpack_social_options":{"image_generator_settings":{"template":"highway","default_image_id":0,"font":"","enabled":false},"version":2},"_wpas_customize_per_network":false},"post_tag":[423982349],"company":[81,171,423954278,423988591],"channel":[186,12374,423875666],"location":[],"nyo_column":[],"person":[423986813],"nyo_post_hidden":[],"coauthor":[423983223],"class_list":{"0":"post-1653092","1":"post","2":"type-post","3":"status-publish","4":"format-standard","5":"has-post-thumbnail","7":"tag-bloomberg-tech-summit","8":"observer_company-google","9":"observer_company-meta","10":"observer_company-openai","11":"observer_company-scale-ai","12":"channel-business","13":"channel-artificial-intelligence","14":"channel-technology","15":"nyo_person-alexandr-wang"},"acf":{"homepage_position":"","homepage_title":"","homepage_excerpt":"","alternative_og_image":"","headline":{"seo_headline":""},"subheadline":{"optimized_seo_description":"","optimized_social_excerpt":""}},"apple_news_notices":[],"parsely":{"version":"1.1.0","canonical_url":"https:\/\/observer.com\/2026\/06\/alexandr-wang-defends-meta-muse-spark-model\/","smart_links":{"inbound":0,"outbound":0},"traffic_boost_suggestions_count":0,"meta":[],"rendered":"","tracker_url":"https:\/\/cdn.parsely.com\/keys\/observer.com\/p.js"},"jetpack_publicize_connections":[],"jetpack_featured_media_url":"https:\/\/observer.com\/wp-content\/uploads\/sites\/2\/2026\/06\/55315389780_2a7308087e_k-e1780684334665.jpg?quality=80","coauthors_byline":"By Victor Dey","display_channel":"","thumbnail":"\"Alexandr