ResearchJul 11, 2026
Anthropic researchers find a "workspace" inside Claude where its reportable thoughts live
New interpretability research from Anthropic identifies a small privileged region of Claude's internals — dubbed J-space — that behaves like a mental workspace for thoughts the model can hold, use, and report.
Read more →
Model ReleasesJul 2, 2026
Claude Sonnet 5 brings near-flagship AI agents down to a mid-tier price
Anthropic released Claude Sonnet 5 on June 30, a mid-priced model that plans, uses tools, and completes multi-step work at a level that previously required its most expensive models.
Read more →
Model ReleasesJun 16, 2026
Anthropic's most powerful model yet launched, then went dark within days
Claude Fable 5 arrived on June 9 as the first model in Anthropic's new Mythos-class tier, but a US government export-control directive forced the company to pull access worldwide just three days later.
Read more →Never miss a useful update
One AI tool and the week's most useful news, every Monday.