Meta launches Muse Wonder agent model, gating it behind $20 and $100 monthly Muse app plans
Meta released Muse Wonder, a model it claims is its most advanced for powering digital assistants. Access requires a $20 or $100 monthly Muse app subscription, and a larger Muse Wonder Full variant is limited to Muse Code subscribers. API pricing has not been disclosed.
Muse Wonder — Quick Specs
Meta released Muse Wonder, a new AI model that the company says is its most advanced option for powering digital assistants, and put it behind a paywall. Users of the Muse personal agent app must subscribe at $20 or $100 per month to access it, according to CNBC, which first reported the launch on October 12, 2026.
What Meta released
Muse Wonder, codenamed "Watermelon," is built to improve what Meta calls "personal and professional agents." Meta AI chief Alexandr Wang said in an interview on Monday that the model was trained to handle real-world tasks such as booking dinner reservations and checking availability.
Wang described the design goal: "When you do real-world tasks, you're going to get roadblocks and you're going to get things that come up or various setbacks. We specifically tried to train the model to be able to try other approaches instead of stalling and then checking back in with the user at any time."
Meta also claims in a blog post that the model "offers our strongest safeguards to date" against "cyber misuse." Those protections currently apply only to the paid model.
Access tiers and pricing
| Offering | Access | Price |
|---|---|---|
| Muse Wonder | Muse app | $20 or $100 per month |
| Muse Wonder | Meta Model API | Not yet disclosed |
| Muse Wonder Full | Muse Code agent | $50 or $200 per month |
- Meta Model API: Wang said API pricing will be higher than the standard subscriptions because Muse Wonder is a "much larger model and at frontier level." He said it will still be priced "meaningfully below" leading models from OpenAI and Anthropic. Per-token pricing has not been published.
- Muse Wonder Full: This more powerful variant is available only to developers who subscribe to Muse Code, Meta's software-engineering-focused agent.
- Muse app: The app itself remains free, with subscription plans for higher usage.
Meta has not disclosed parameter count, context window, training cutoff, or benchmark results for either model. The claim that Muse Wonder is Meta's most advanced assistant model comes from the company and has not been independently verified.
Muse app incidents
The Muse app launched last month and rose to the top of Apple's App Store. Some users have since complained on social media that the agent accessed personal messages without authorization. Meta disputes this.
Wang said the AI team is monitoring for agent-related "incidents or issues." He suggested some cases involve users who authorized communications earlier in a conversation and later forgot. Meta is trying to better educate users on how the app works "to ensure that consumers can trust the product going forward."
What this means
Meta is using Muse Wonder to test whether consumers and developers will pay for agent capability, a break from a business built mainly on advertising. Gating the strongest safeguards behind the paid tier is a notable choice. Free-tier users get the less protected model, and that will draw scrutiny given the unresolved complaints about unauthorized message access.
The tiering also shows Meta segmenting its lineup: consumer agents, a frontier-priced API, and a top-end Full variant reserved for coding. The API price will be the key number for developers. Meta's "meaningfully below" OpenAI and Anthropic positioning is a claim until per-token rates, context limits, and benchmarks are published.
Related Articles
Cloudflare releases Clef decision models, claims 39 ms median latency vs. 524 ms for TypeSafe's Jev
Cloudflare has released Clef and Clef-flash, two open-weight decision models that return probabilities over predefined answer options instead of generating text. The company claims median latencies of 39 ms and 209 ms, against just over 524 ms for TypeSafe AI's Jev. Both support text and images and are API-compatible with Jev.
Odyssey-3 world model enters free public preview; 14B base model, 66.1 Physics-IQ claim carries caveats
Odyssey has opened a free public research preview of Odyssey-3, an autoregressive diffusion transformer that generates interactive environments from text prompts in real time. The base model has 14 billion parameters and outputs 832 × 480 video. Odyssey's headline Physics-IQ score of 66.1 does not meet the benchmark's own requirements for a record claim.
Microsoft releases Decision-1, a Qwen3.5-9B-based model for classification and routing, at $0.042 per 1M input tokens
Microsoft has released Decision-1, a decision model built on Qwen3.5-9B for classification, evaluation, and routing. Microsoft claims 83.5% accuracy across 36 benchmarks and 85 ms latency. Input tokens cost $0.042 per million, and output tokens are free.
Microsoft releases FrogNano-4B, an Apache 2.0 coding agent trained with RL on 1,500 synthetic tasks
Microsoft has released FrogNano-4B-2609, a repository-level coding agent derived from Qwen3.5-4B and published under Apache 2.0 with open weights. Microsoft says it was post-trained only with reinforcement learning on about 1,500 synthetic software-engineering tasks, with no stronger-model trajectories. It is evaluated at roughly 131K tokens of context.
Comments
Loading...