Claude Fable 5.1

“Fable 5.1 and Astra working together beats each model on its own”

“Better than hitting straight walls with Anthropic's Opus 5 or Fable 5.1 under Claude Max 5x”

“Fable shouldn't be coding. It should be researching and planning.”

“Fable 5.1 is a token hog”

“Claude Fable 5.1 went 3 hours in loop earlier”

“They fixed the safeguards last week”

“Fable 5.1's release page says it's better at these perspectives of coding and per my experience, yes it is”

“Using Fable's story elements could definitely help with understanding programming concepts”

“for code reviews and raw coding, Fable 5.1 is just glorious to work with”

“I was switching between GPT-5.6 Sol and Fable 5.1 to check each other's work in Cursor. And the result was gradually driving me insane. As the models struggled to find a solution that would actually work, they dug themselves deeper into a h”

“SuperTuxKart is in the training data. Recreating existing software is not that impressive, nor a good benchmark.”

“I'm on like phase 60+ with multiple sub phases on mine but also I feel like I haven't really approached anything resembling "done" on my little hobby game.”

“Fable 5.1 has done MASSIVELY better in regards to usage consumption then 5.0 did”

“5.1 is very good at accomplishing tasks, but not as good at explaining how the task was accomplished, or complex topics in general, it's more terse”

“needs a fairly high level of human guidance”

“Mere mention of "reverse engineering" gets me kicked back to Opus.”

“It refused to answer anything about pickling for me. Always kicked me down to Opus which is comical.”

“Real testing shows fable-5.1's token efficiency is noticeably higher, saving 1/3 the cost compared to the previous fable5, and now it also feels faster.”

“Cache zone token costs decreased by 75%, bringing overall usage costs down by nearly half.”

“if I ask it to paint with a shade of red, but it paints with a slightly different shade of red, that is a fucking effect on quality”

“The safety filter is simply no longer an issue.”

“Claude Fable 5.1 also has its own strengths. In HLE tests covering multi-disciplinary difficult problems with tool use allowed, it scored 65.0%, compared to Astra's 57.2%; AA comprehensive intelligence indices were 65.7 and 61.2 respectively.”