OMITTED

What the news leaves out.

← Omitted front page

Meta Launches New Muse Wonder AI Models

6 sources · updated 2026-10-13
Left 67% Center 33% Right 0%
4 left · 2 center · 0 right

What happened

On Monday, Meta’s Superintelligence Lab released two new AI models, Muse Wonder and Muse Wonder Full, which the company described as its first frontier models. Muse Wonder is available to Muse Code subscribers and through the Meta Model API at $4 per million input tokens and $20 per million output tokens; Muse Wonder Full is limited to Muse Code subscribers and is not served via the API. Meta said both models will be used to power its Muse AI agent and, based on its own testing, posted benchmark scores comparable to OpenAI’s GPT-6 Astra and Anthropic’s Claude Fable 5.1, with particular emphasis on an “active memory” capability. Access for Muse app users is tied to paid plans; Muse Wonder Full is offered under Muse Code subscriptions. The models were trained at Meta’s Prometheus data center in central Ohio.
Sourcing check · Mostly on the record · 6 articles

4 of 5 articles cite a named source for the story. It was on the record from the first report.

What these terms mean and how they're counted
On the record · names a source 4 of 5
On the record means the article cites a source by name for the story's main claim: a person, or an institution speaking officially, such as a company statement or a named agency's spokesperson. The opposite is an unnamed source. The count is how many of the counted articles cite at least one such source; the names are everyone named across them, each once however the articles spell it. Here: Meta, Alexandr Wang, Mark Zuckerberg and Kavya Kopparapu
Primary evidence · cites a document or data 1 of 5
Primary evidence means the article cites a record you could check yourself: a court filing or ruling, a dataset or report, a recording or transcript, an official statement or press release, market data. A quote is not a document. The count is how many articles cite one. Here: model card, blog post
Credits another outlet 0 of 5
The article says the story comes from another outlet's reporting (“first reported by Reuters”). Those articles relay the story; they are not independent confirmation. Here: Each article presents its own reporting.
Unnamed sources 1 of 5
The main claim is attributed to people who are not identified: “people familiar with the matter”, “officials said”. The count is how many articles do this.
How it developed
When the first report landed and whether it named anyone; when the first named source and the first document appeared. Here: First report Oct 12, 17:00 UTC by The Verge, naming its source. First document within the hour, from TechCrunch (model card).
Not counted 1
Opinion, analysis and video pieces, and articles too short to assess. Every count above is out of the counted articles. An article republished under a new headline is one article.
This is about who is cited, not whether the story is true. Each source below says what it names. How we count
ONE-SIDED THIS WEEK. Only left-leaning outlets have covered this story in the last two days; Fox Business reported it on 2026-09-29.
Omitted — what each side leaves out

Unpacked

Right-leaning outlets had not covered this as of publication. Left-leaning coverage reports that Meta launched Muse Wonder and Muse Wonder Full and claims they can rival OpenAI and Anthropic, but it splits between a catch-up story and a paywall story. CNBC alone reports that Meta's "strongest safeguards to date" against cyber misuse come with the paid model for now. A reader of the silent side misses the release, and that the best safety claim is currently a paid feature. CNBC opens on a test of whether users will pay, putting a $20 monthly plan in the lede, while The Verge opens on Meta competing with the industry's best and never names a price. Does any lab besides Meta confirm that Wonder matches GPT-6 Astra?
Bottom line

Depending on the left-leaning outlet, Meta either just rivaled GPT-6 Astra or just put Wonder behind a $20 paywall, and right-leaning readers were told neither.

The receipts — 5 facts one side's coverage lacks
  • Meta said the latest model offers its strongest safeguards to date against cyber misuse, and those protections only come with the paid model for now official statementin CNBC · absent from right-leaning coverage
  • Accessing Muse Wonder in the Muse personal agent app requires a monthly subscription of $20 or $100 numberin CNBC · absent from right-leaning coverage
  • Muse Wonder Full is limited to Muse Code subscribers on a $50 or $200 a month plan numberin CNBC · absent from right-leaning coverage
  • Some users complained that the Muse agent accessed personal messages without authorization, which Meta disputed individual casein CNBC · absent from right-leaning coverage
  • Through the Meta Model API, Wonder costs $4 per million input tokens and $20 per million output tokens numberin TechCrunch · absent from right-leaning coverage
The Left View
Left-leaning outlets including The Verge, The New York Times, CNBC, and Business Insider framed the launch as Meta trying to reenter the frontier-model race with OpenAI and Anthropic after investing heavily in talent and infrastructure. They reported Meta’s claims that Wonder and Wonder Full rival GPT-6 Astra, Claude Opus 5.5, and Claude Fable 5.1; subscription prices of $20 or $100 a month for Muse and $50 or $200 for Muse Code; CEO Mark Zuckerberg’s statement that the series brings Meta “to the frontier”; and AI chief Alexandr Wang’s comments that the models (internally codenamed Watermelon) were trained to retry real-world agent tasks instead of stalling. CNBC noted Meta’s claim of stronger “cyber misuse” safeguards on the paid model, user complaints that the Muse agent accessed personal messages (which Meta disputed), and the company’s effort to build AI revenue beyond ads. Center outlets TechCrunch and CBS News covered the product details, API pricing, MemoryBench results as reported by Meta, and the models’ role as the first frontier release since Wang restructured Meta’s AI work.
Our Take (balanced)
The story is substantive as a corporate product announcement: named models, prices, availability, a model card, a company blog post, and on-the-record comments from Mark Zuckerberg and Alexandr Wang are specific and checkable. Established facts are that Meta released Muse Wonder and Muse Wonder Full, set the stated prices and access limits, and claims internal benchmarks comparable to leading rival models, including higher MemoryBench scores. Those performance claims remain Meta’s own, uncorroborated so far by independent third-party tests in this coverage. Reports that the Muse agent accessed personal messages without authorization are allegations Meta has disputed, not proven facts. No right-leaning outlet has reported this. Watch for independent benchmark results, API documentation, whether Wonder actually powers the Muse agent as promised, and any official follow-up on the privacy complaints.

6 sources

More stories

The week's bottom lines, in your inbox

One email a week: the five stories that mattered and what they actually mean. Free.