We Tested Fable 5.2 Early. It’s a BEAST.
Dan Shipper of Every reviews Claude Fable 5.2 after early testing. He rates it S tier overall and calls it a paradigm shift for coding and some knowledge work, though weaker for writing.
The gist
-
Tyler Nishida calls it an overnight model that gives ETAs; Klaassen prefers it for hand-offs and Opus 5.5 for staying in the loop.
-
Shipper says it is the first model whose games are actually fun, turning a photo of toys into a Mario Kart style racer.
-
Mike Taylor had it turn his research into a documentary, using about 1.4 billion tokens, nearly $2,000 at Fable 5.1 rates.
-
Shipper rates writing B tier, preferring Gemini 4 Argon, but design A tier, citing details like a young Socrates with a black beard.
Also mentioned
2:41
EveryCut
· "This is how we cut video now. It has an agent built in. It can transcribe everything it sees. I can record..."
3:05
Premiere
· "The really interesting thing about EveryCut is not just that it's got all the features of Premiere, it's..."
3:14
Descript
· "It's got features of all these other platforms that we need internally at Every, like Descript and..."
3:14
Streamyard
· "It's got features of all these other platforms that we need internally at Every, like Descript and..."
◎
Related episodes
Episodes covering similar topics
More from Every
A New Vision for Human Work After Automation
Nov 11, 2026
LIVE Vibe Check: Gemini 4 Carbon
Nov 11, 2026
Fable 5.2 Vibe Check
Nov 10, 2026
Every Think Week: Monologue HARDWARE?!
Nov 03, 2026