Something worth paying attention to in the Fable 5 launch that I think will get buried under benchmark comparisons. The most consequential line in the AWS announcement wasn’t about context windows or coding performance, it was tucked into the infrastructure section: “Once you opt into data retention, your data will leave AWS’s data and security boundary.” That’s not a model feature, that’s an enterprise architecture constraint. For a lot of companies that sentence alone disqualifies Fable 5 from touching certain workloads no matter how good the model is.
The Fable vs Mythos split is also worth sitting with. Same underlying capability apparently, but Mythos is gated behind Project Glasswing and vetted partners only. Anthropic is essentially saying some capability is too sensitive for flat API access, which is a pretty different philosophy than “here’s our best model, go build.”
Does the Fable/Mythos split read as responsible deployment to people here or more like managed scarcity? And anyone in enterprise AI already hitting the retention requirement as an actual blocker?
[link] [comments]