BREAKING
Claude Fable 5 Redeployed Globally
0
M
token context
0
%
SWE-bench Verified
0
K
max output
SWE-bench Pro Results
Fable 5
80.3
Opus 4.8
69.2
GPT-5.5
58.6
0
$
per M input
0
$
per M output
Where to Use Fable 5
Use For
Best fit
●
Complex refactoring
●
Long-horizon agent tasks
●
High-difficulty reasoning
Avoid
Wasteful
●
Plain code generation
●
Short text drafting
●
Burns caps fast
Power Meets Tight Usage Caps
AI NEWS BLITZ
Anthropic's top-tier Claude Fable 5 is back online after an export-control pause.