Normally when we test a new model, we’re looking at how far we can push it. Claude Fable 5, released today, is the first model we’ve tested that felt like it was pushing us.

The Every team spent five days handing Fable ambitious projects across coding, writing, knowledge work, and custom agents. Our day-zero Vibe Check covers:

• Where Fable dramatically outperformed every model we’ve tested
• Who is well-positioned to take advantage of the model’s capabilities
• When to reach for Fable, and when a faster, cheaper model will serve you better
• Coming Wednesday: Dan Shipper’s podcast with Mike Krieger, the head of Anthropic Labs, on how Fable changes work

​Paid Every members also get:

• Our favorite Fable use cases and how to try them yourself
• Our team’s reach test for coding, writing, knowledge work, and agent behavior
• An inside look at the results from our benchmarks for coding, writing, and general knowledge work
• Access to our power user camps this month for Fable and Codex​

​Start your free trial today to unlock it all.