Fable

“A single workflow might use Fable 5 for one step, GPT-5.6 Sol for another and Kimi K3 for a third”

“when Fable 5.1 was released, the Anthropic-gets-to-keep-your-data provision was gone”

“customers pushed back, and Fable usage stayed relatively low”

“Fable is fantastic. Like you're going to give it a really complex like engineering task. I'm like, great. Have Fable go take a swing at this. It's going to nail the thing. But man, I do not want to talk to Fable.”

“Fable is a wonderful... I think that's probably the best visual taste right now. It feels like it actually put some thought into its decisions compared to previous models.”

“Same $5 in, $25 out as Opus 4.8. Half of what Fable 5 charges.”

“I spent the last week and ≈$2k building a pretty slick replacement for a $10k SaaS product with Fable while consolidating together a bunch of vibe coded side projects that I'd built to supplement the SaaS we're now churning from.”

“Claude Opus 5 matches Fable 5 on coding benchmarks at half the cost”

“Claude Opus 5 matches Fable 5 on coding benchmarks at half the cost”

“Fable's own launch video was edited entirely by Claude Code, combining transcription, ffmpeg cuts, Remotion prototypes, and Claude-taught color grading.”

“"The first genuine (and fastest) megakernel ever submitted to KernelBench-Mega, according to one of the benchmarks maintainers."”

“Fable achieved an 18.71X speedup vs. an optimized PyTorch baseline, compared to 14.4X (Claude Opus 4.8) and 4.34X (GPT 5.5) — and did so with a single cooperative kernel launch per decoded token, vs. 4–14 launches for all other high-scoring entries.”

“Fable achieved an 18.71X speedup vs. an optimized PyTorch baseline, compared to 14.4X (Claude Opus 4.8) and 4.34X (GPT 5.5) — and did so with a single cooperative kernel launch per decoded token, vs. 4–14 launches for all other high-scoring entries.”