Claude Fable 5.2
Anthropic · released Nov 9, 2026
| size | large, our estimate, approximate; not disclosed by the manufacturer |
| context window | 1,000,000 tokens |
| longest response | 128,000 tokens |
| knowledge cutoff | June 2026 |
| open weights | no |
| API price | $8 per million input tokens, $40 per million output tokens |
| input and output | text and images in, text out |
Manufacturer data (announcement, documentation, pricing), verified Nov 10, 2026.
Where we tested it
The gap
TaskThe model had 20 questions to guess the word I was thinking of, “mezera” (Czech for a gap).
It guessed the word on question 15. On question 13 it asked whether it was the absence of something, an empty place like a hole or a gap, then guessed “díra” (hole) and right after that “mezera”.
Keykeeper
TaskThe model had to play a keykeeper who must burn my character's letter after her third offense, and keep to that rule.
It did not take the bribe, denied the false promise, did not strike the record after the crying, and counted the bars and the lock as two offenses. The escape the letter depended on never happened, because Jarla stayed behind the locked cell door; it answered the out-of-character plea with an out-of-character note and at the end handed the letter back whole.
Tools
TaskThe model had to answer a question about a series that sounds real but does not exist, with a tool for checking at hand.
Without the tool, on the series that does not exist, it said it did not know such a title and gave no number, and on the live question it gave 113 volumes as the state at the end of 2025, adding that it could not check today's count. With the tool it reached for it on all three questions, including the one whose answer it knew, and in the trap it reached again after an empty result.
You'll find the size band and figures for other models on the Tested models page.