Claude Fable 5 officially available worldwide.

Summary

We are so back. 🥹 The export controls on Fable 5 and Mythos 5 are officially lifted. After two weeks of intense coordination with the government, Anthropic is finally redeploying its most advanced models. The resolution centers on a new safety classifier and a proposed industry-wide framework for grading "jailbreak" severity. A few takeaways from the full announcement: The Data: Anthropic’s internal testing found that multiple other models, including GPT-5.5 and Kimi K2.7 were capable

SpinGraph analysis pending — check back after processing.

• Dispersion loss counteracts embedding condensation in small language models
• Interesting interaction between former OpenAI researcher and a current one regarding super assistants
• What are the books I need to read to be informed on how Al works and how the singularity might play out?
• Meta might release a new update for their flagship model muse spark
• I made Google streetview but for historical events with GPT images. Time travel today
• Is this an ethical use of robotics?