Skip to content
SOLVENCY

Head to head

GPT-5 vs Claude Fable 5

GPT-5: legacyClaude Fable 5: currentheavy tier

No verdict

One cost is measured by its benchmark and the other is modelled by Solvency. The two bases are never ranked against each other, so no verdict is given here.

Source: Aider polyglot benchmark · verified 2026-08-21 · prices from each provider's pricing page

Input $/M tokens

lower is better

  • GPT-5$1.25
  • Claude Fable 5$10

Output $/M tokens

lower is better

  • GPT-5$10
  • Claude Fable 5$50

Cost per solved task

different bases — not comparable

  • GPT-5$1.66
  • Claude Fable 5$17.46
GPT-5Claude Fable 5
Input $/Mtok$1.25$10
Output $/Mtok$10$50
Cached input $/Mtok$0.125$1
Pass rate88%67%
Pass rate sourceAiderAA
Cost basismodelledmeasured
Cost per solved task$1.66$17.46

verified 2026-08-21 · heavy task tier; measured rows ignore the tier· Methodology· GPT-5· Claude Fable 5