Many users praised Fable as more efficient than GPT-5.6 with less babysitting needed and welcomed the competition, while others called it unusable due to quick limit exhaustion and high cost.
Based on 16 visible X reactions from 32 accounts; directional sample.
Ask a question below.
Published answers will appear here.
Refreshing to see an honest take. Some might view this as negative review but to me, it’s a balanced one. This does your perception of trustworthiness a lot of good since when the ‘GPT guy’ offers a genuine take on the model and how it compares with Fable, it’s not just “GPT is always the best”.
@mattshumer_ wild to see that Anslopic actually delivered something that much better
@mattshumer_ RL stack is the real secret sauce here. Can't wait to see that beast in action.
@mattshumer_ Fable is unusable; the limits run out immediately. You can't work with it
@mattshumer_ Efficiency... You did not rate it. OpenAI is a beast, Anthropic is poor.
@mattshumer_ Fable-5 is the king right now, gpt-5.6 is the second choice
I've had access to GPT-5.6 since May 27. Everyone is going to tell you it's incredible. They're right. But this isn't another post glazing 5.6. For 2 weeks, it was the best model I'd ever used. Then Fable came out, and I stopped using GPT-5.6 overnight. Here's my review: https://twitter.com/mattshumer_/status/2075267645058805896
Most people know I usually prefer OpenAI's models over Anthropic's. But for this generation, it's the opposite. That said, a larger pretrained base + OpenAI's RL stack will create an absolute beast of a model, so I expect OpenAI's next release will bring me back. https://twitter.com/mattshumer_/status/2075267773458940306
Many users praised Fable as more efficient than GPT-5.6 with less babysitting needed and welcomed the competition, while others called it unusable due to quick limit exhaustion and high cost.
Based on 16 visible X reactions from 32 accounts; directional sample.
Ask a question below.
Published answers will appear here.
@mattshumer_ Fable-5 is the king right now, gpt-5.6 is the second choice