TLDRocket
Sign in

Zuck’s wonder...full

Ben's Bites

Opinion — commentary, not a factual news event.

Meta launched Wonder and Wonder Full, two new AI models. They’re cheaper than rivals, but the real fight is over who gets builders to use them.

Based on reporting by Ben's Bites — read the original for the full story.

Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error

Ben’s Bites is doing double duty here: half rant, half product roundup. The rant says the old idea of software is getting bent out of shape by personal agents, headless tools, and people who would rather ask a bot than open an app. The product list is the proof pile underneath it.

Meta’s pair, Wonder and Wonder Full, sit at the center. Wonder is pitched as a cheaper Fable/Astra-class model, priced at $4 and $20 per million tokens versus $10 and $50 for Fable and Astra. It’s available through Meta’s API and in Muse Code, and Meta says it will also power Muse, the company’s personal agent. Wonder Full is the more capable version, described as just shy of Opus 5.5 everyday performance, but for now it lives only in the Muse Code subscriber plan, which starts at $50 a month.

The catch, at least in this first pass, is that the models didn’t exactly inspire love. Ben says Wonder Medium was chattier than Wonder Full High, both felt slow, and both produced awful UI. He didn’t even bother testing them on proper coding work, which is a pretty blunt verdict for something that’s supposed to sit near the frontier.

Elsewhere, Claude Code Projects is now open to every Pro and Max user, giving people one master chat to dump requirements into while Claude spins off separate threads. Cursor, Grok, and X now have a four-tier plan stretching from $8 a month to $200 a month, with more usage for Grok and Composer models but tighter limits on Claude and GPT. Grok Bot also has its own email address now, because apparently even bots need inboxes.

The broader theme is very Ben’s Bites: stop treating software like a polished shrine and start treating it like something bots can poke at, break, and file away when they’re done. That’s where the money is going, and that’s why the interfaces keep getting pushed aside.

My take — AI-written commentary, not fact-checked reporting

The industry keeps pretending agents are a side feature when they’re clearly becoming the front door. That’s bad news for anyone selling glossy dashboards and good news for people who think an app should mainly get out of the way. The new status symbol is software nobody has to look at twice.

Read more about this at: Ben's Bites

Related stories

The daily briefing

Every AI story that matters, in your inbox by 8am.

TLDRocket reads all relevant sources, removes duplicate coverage, and summarises the day in two minutes. Follow companies and topics for alerts, or get the briefing in Slack. Free, no spam, unsubscribe anytime.