Claude Fable 5
Model at Anthropic
The basics
As of Nov 1- Status
- Price
- Context
- Open weights
Updates
Role
Long-horizon reasoning model at Anthropic
Sep 2026 to now · 2 months
Jun 2026 to Sep 2026 · 3 months
Works with
From recorded changes
App · 6 shared changes
Agent · 6 shared changes
API · 5 shared changes
Agent · 5 shared changes
Similar models
$10 in, $50 out
Cost vs peers
Tied for cheapest of 2 similar models by output price.
Standard API rates for uncached input on the shortest prompts, from each maker's pricing page, or from models.dev for models not researched here yet.
Where it works
Prices and uptime are OpenRouter's figures for each provider's cheapest endpoint. Uptime covers the last 24 hours.
Benchmarks
Higher is better. Green is Claude Fable 5. Dark marks are similar models.
Independent
What people say
Reviews (1)
Lior GrossmanShipping software since 1993, now building with AIFable 5 was the strongest model I had in the summer of 2026, and the most expensive. From June to August I used it for broad, open-ended work: strategy reviews across my products, landing page reviews and planning. It's amazing at connecting the dots across wide issues, and it has great empathy and copywriting. It works best when it agrees on high-level decisions before moving to details, and when it asks me questions before deep research, so it doesn't make things up. What did not work: it does more than asked. When I told it to change one file, it took that as permission to change others. It answered in technical jargon and dense responses I didn't always understand. When Cowork's Default setting moved my scheduled jobs to Fable 5, my usage tripled overnight.
Oct 30, 2026
Launch quotes, picked by Anthropic
From its launch post, so not a neutral sample.
"Claude Fable 5 is the state of the art model on CursorBench. It's opened up a class of long-horizon problems that were out of reach for earlier models."
"Claude Fable 5 is a real step forward for the developers GitHub serves. In our early testing, it took on complex, long-horizon coding tasks with a level of autonomy and reliability that exceeded previous benchmarks. But what excites us most is the direction it points: a future where developers can hand increasingly ambitious work to agents and trust the results across the software lifecycle."
Incidents
Incidents per week, last 13 weeks. Most in one week: 2.
- Oct 24Anthropic reported elevated errors for Claude Fable 5 and Fable 5.135 min
- Oct 21Anthropic reported slow performance for Claude Fable 540 min
- Sep 22Anthropic reported elevated errors across several Claude models1 h 20 min
- Sep 3Anthropic reported elevated errors across several Claude models2 h 57 min
- Aug 24Anthropic reported elevated errors on Claude models3 h 24 min
Writing habits
Creative writing · EQ-Bench, among 51 models tracked here
Slop score
10.3
Overused AI words and phrases. Less slop than 82% of tracked models.
Repetition
3.9
Repeated words and ideas. Less repetition than 46% of tracked models.
Story length
5,887
Average characters per story. Longer than 24% of tracked models.