Claude Fable 5 Launched June 9 as Most Capable Public Model Before Shutdown
Fable 5, released June 9, was positioned as a guardrailed sibling of Mythos 5 that blocks high-risk cyber and bio responses, and posted a field-leading 80.3% on SWE-Bench Pro. Anthropic touted a defense-in-depth safety strategy plus 30-day data retention to detect and mitigate jailbreaks. Three days later the US directive took it offline, making its market life one of the shortest for a frontier model.
Fable 5, released June 9, was positioned as a guardrailed sibling of Mythos 5 that blocks high-risk cyber and bio responses, and posted a field-leading 80.3% on SWE-Bench Pro. Anthropic touted a defense-in-depth safety strategy plus 30-day data retention to detect and mitigate jailbreaks. Three days later the US directive took it offline, making its market life one of the shortest for a frontier model.
Source: Build Fast with AI · 2026-06-12 · https://www.buildfastwithai.com/blogs/ai-news-today-june-12-2026
Related Articles
Magnificent Seven Sheds $2 Trillion in June as AI IPO Wave Spooks Markets
Microsoft, Amazon, Apple, Alphabet, Nvidia, Tesla and Meta have collectively lost about $2 trillion in market value in June, more than two-thirds of the S&P 500's total loss, on fears the SpaceX, Anthropic and OpenAI listings will pull capital from incumbents. Goldman Sachs argues the selloff is overdone given ~$8 trillion parked in money-market funds.
Claude Code Adds Nested Sub-Agents and Plugin Search in Major Update
Anthropic shipped a significant Claude Code update this week adding nested sub-agents - a primary session that spawns specialized agents to run tests, write docs and review code in parallel through a shared task graph - plus smarter model/region selection, a plugin search interface, and tighter Chrome, VS Code and terminal integration. The release also fixes a wide range of session, memory and permissions issues.
Apple's Private AI Cloud Runs on NVIDIA GPUs Inside Google's Data Centers
Apple confirmed its Apple Foundation Models on Cloud - the private infrastructure behind Siri AI's reasoning - runs on NVIDIA GPUs hosted in Google's cloud, with the top-tier AFM Cloud Pro said to match Gemini Frontier quality. Apple is effectively renting both model and hardware from rivals while wrapping it in its Private Cloud Compute privacy guarantees.