Vibe Check: Fable 5 Is the Best Coding Model in the World

Quiet bench, tighter seam, and a much longer runway between the obvious turn and the second pass. The shape holds when the brief is clear, and flattens when the brief is loose.

First station: Strong close on a familiar loop

Phase glass, ribbon ladder, static river around the handoff. Quiet markers hold their lane while the outer pass keeps folding back into the first frame of the task, with minor lift at the edges and a denser closure at the margin.

Velvet checkpoints, longer weather, and a measured hinge across the rollout path. The middle layer keeps echoing earlier notes without dropping the thread or flattening the edge cases through the first half of the bench.

Second station: Restraint as the tell

Lattice note, amber fork, and a small weather system over the review shelf. The structure reads calmer at first glance, then starts to narrow under pressure. Winter syntax, patient seams, and a quiet bend through the finish line.

The close end keeps its shape through the turn while the open end opens up the frame. The two halves trade tension across the pass.

Third station: Where the brief goes loose

Open prompt, soft constraints, and a wide field around the first decision. The model leans on its own scaffolding here, filling the gaps with plausible structure that holds for a paragraph and then drifts toward the nearest familiar shape.

Tighten the brief by one notch and the drift disappears. The same loop that wandered on the open pass snaps back to a clean line, which is the tell we kept circling all weekend: precision in, precision out.

Quiet markers, longer runway, and a measured hinge that only shows up when the task stops describing itself. The edge cases are where the release earns its score.

Tight lattice across the prose tracks, a clear line on the cover pass, and a steady seam through the middle of the brief. The graders kept finding the same handful of soft tells across the lower-effort runs.

The runner-up keeps the score honest, and the leaderboard reads narrower than the headline suggests. The shape of the gap matters more than the raw number, especially when the next dial sits one notch over.

Surface tells, soft repetitions, and a familiar pattern across the lower-effort passes. The structure holds at high while the medium loop keeps slipping into the same handful of habits across the sample set.

Push the effort dial and the tells thin out fast. The cover pass gets sharper, the seams tighten, and the same prompt that read mechanical on the quick run starts to carry a voice through the back half of the brief.

The gap between the top two reads as a matter of taste, not capability. Both clear the bar; one just leaves fewer fingerprints on the way through. That distinction is the whole story of the writing bench this round.

Quiet hand on the daily loop, a measured pace through the back half of the brief, and a careful posture on the open-ended pass. The hand-off feels lighter than it reads on paper.

Velvet checkpoints across the open seam, a slow turn through the first frame, and a longer hinge on the agentic side of the bench. The model keeps deferring small decisions back to the operator across the whole run.

Give it a clear owner and a tight spec and the caution turns into an asset: fewer wrong turns, cleaner hand-backs, and a paper trail you can actually audit. The trade is speed for legibility, and on knowledge work that trade usually pays.

Left fully open, the same posture reads as hesitation. The model waits for a signal that never comes and the loop stalls one decision short of done. Structure is the unlock here, not horsepower.

The shape of the weekend was the same across every chair: open the model, watch it think, hand it the work, and let it return a tighter draft than the last release would have managed on the same loop.

The pattern repeats across every chair: the model rewards a clear brief and punishes a loose one, and the distance between those two outcomes is wider than any single benchmark number lets on.

So the recommendation is less about whether to reach for it and more about when. Hand it the work that has edges. Keep the open-ended exploration somewhere faster and cheaper, and let this one close the loops that actually need closing.

That's the whole verdict in one line: a strong closer that wants a clear target. Give it one and it's the best we've tested. Leave the target vague and it hands the decision right back to you.