Claude Fable 5 and Claude Mythos 5 remained unavailable globally on June 22, entering a tenth day of suspension following the Trump administration’s June 12 export control directive that required Anthropic to block access by any foreign national, inside or outside the United States (Anthropic). Rather than attempting nationality-based filtering, Anthropic disabled both models for all customers worldwide (Quartz), citing a reported jailbreak method capable of bypassing Fable 5’s cybersecurity guardrails (Anthropic). Co-founder and chief compute officer Tom Brown and policy chief Sarah Heck were dispatched to Washington to negotiate with White House officials, and Anthropic confirmed President Trump eased national security concerns after meeting CEO Dario Amodei at the G7 summit in Evian-les-Bains (TechTimes). Anthropic’s international managing director stated on June 18 that the company was “very confident” both models would return “in the coming days” (TechTimes), though prediction markets as of June 22 place 57% odds on restoration before July 1 and 75% by July 17 (Kalshi). June 22 marks the last day Fable 5 is included in Pro, Max, Team, and Enterprise paid plans at no extra cost, per Anthropic’s announced transition timeline (Anthropic).
In other developments, Z.ai (formerly Zhipu AI) released GLM-5.2 model weights under the MIT license on June 16, hosting them at zai-org/GLM-5.2 on Hugging Face (Simon Willison). The 753-billion-parameter mixture-of-experts model runs 40 billion active parameters, supports a 1-million-token context window, and scored 62.1 on SWE-bench Pro - ahead of GPT-5.5 at 58.6 and GLM-5.1 at 58.4 - though security researchers note that API usage routes data through Z.ai’s Chinese infrastructure (TechTimes). Separately, xAI’s Grok 4.3 became generally available on Amazon Bedrock on June 15, priced at $1.25 per million input tokens and $2.50 per million output tokens, making it the lowest-priced US-lab frontier reasoning model on the platform (AWS); the model features a 1-million-token context window and configurable reasoning effort levels (none, low, medium, high), running on Amazon’s new Mantle inference engine with support for tool calling, structured output, and response streaming (xAI).