Sonnet 5 vs Fable 5

A meaningful capability and price gap, not a marginal one

Fable 5 sits well above Sonnet 5 on price, reflecting a real capability gap rather than a small premium for marginal improvement — this is the comparison to make when a task genuinely might need the top tier, not a default pairing to weigh for routine work, where Sonnet 5's much lower price and strong coding and agentic performance make it the more sensible starting point.

Thinking behaviour differs in a way that matters beyond price

Fable 5 runs thinking on every request with no way to disable it; Sonnet 5 defaults to thinking on but retains more flexibility around it depending on effort level. For latency-sensitive work where even Sonnet 5's default thinking overhead is unwelcome, that's a further consideration beyond raw capability or price — Fable 5 offers no path to reduce that overhead at all.

Why Opus 5 usually belongs in this conversation too

In practice, a team comparing these two specifically is often missing the more relevant middle option — Opus 5 frequently closes most of the gap to Fable 5's capability at a meaningfully lower price than Fable 5 itself, which makes a direct Sonnet-versus-Fable comparison less useful on its own than weighing all three together.

Related

See Opus 5 vs Fable 5 for the more common comparison against Fable 5's actual practical alternative, and the model picker for a recommendation weighing all three factors — price, capability, and thinking behaviour — against your specific task.