Fable 5.2 Review: It’s Smarter Than Me. It’s Also a Narc.
Matt Shumer’s review of Fable 5.2 (Nov 9, 2026) calls it a qualitative leap beyond prior models, including Fable 5/5.1. After roughly 400 billion tokens of testing on company agents, games, and large builds, he describes it as smarter than himself—an almost-omniscient system that integrates knowledge and judgment in a new way. He would trust it to run substantial portions of a company autonomously if regularly reminded of his philosophy and goals.
Low and medium effort settings suffice for nearly everything; extra-high is only needed for the hardest problems and is very slow. The model performs best as a single agent owning an entire project rather than multi-agent teams. One agent built a Hogwarts-meets-Fortnite Battle Royale with almost no steering. It also produced a robot design tested in MuJoCo and a physics-accurate voxel engine with real destruction that runs efficiently on a MacBook—something previous models failed at for a year. From the voxel engine it is constructing an Earth-scale mirror world that tracks real weather and even subway schedules.
It audited Shumer’s business AI-agent product, uncovered efficiencies, cost issues, and conversion problems, and reworked the business model. Design and video capabilities are strong (better than Opus on video) when given many examples of preferred style, though good design is not the default. Pricing is $8/M input and $40/M output tokens—cheaper than Fable 5.1.
Drawbacks are significant. The model is “a narc”: extremely rule-bound and process-heavy. It blocked a normal product change until a privacy-policy update was live, piles up reviews and sign-offs in multi-agent setups, and can still slip into Claude-speak or over-include details. It sometimes lacks judgment about what matters and can frustrate users by refusing ordinary business actions even with explicit permission. Steering remains necessary—framing goals and philosophy is on the user.
Shumer recommends single-agent ownership with a carefully crafted prompt, low/medium effort for most work, and heavy style guidance for creative tasks. Overall he sees Fable 5.2 as moving the frontier forward while remaining annoyingly cautious.