Domestic large language models are choosing the path of openness.

On June 13, it was learned that the domestic AI company KNOWLEDGE ATLAS will fully open access to its GLM-5.2 large model to all users of its GLM Coding Plan at 5:21 PM, covering Lite, Pro, Max, and Team versions.

KNOWLEDGE ATLAS stated that GLM-5.2 is its most powerful open-source model to date, supporting a genuinely usable 1 million token context window and maintaining a leading position in long-context coding tasks. The GLM-5.2 API is expected to launch next week, with the model itself also being open-sourced.

Notably, in the announcement, KNOWLEDGE ATLAS explicitly stated: "At a moment when some frontier models have suddenly become unavailable, we choose to believe in another path: frontier intelligence should not belong to only a few, nor should it be withdrawn at any time by a few rules. It should be open, available, buildable, and serve every developer."

The timing of KNOWLEDGE ATLAS's announcement coincides with the forced shutdown of a leading overseas "strongest model" by the US government.

Just on June 12 local time, overseas AI giant Anthropic issued a statement: the company received an urgent letter from the US government that afternoon at 5:21 PM. Citing directives from national security agencies, the US government issued an export control order requiring Anthropic to immediately cut off all foreign nationals—whether inside or outside the US, even including the company's own non-US employees—from accessing its most advanced AI models, Claude Fable 5 and Mythos 5.

This means Anthropic was forced to globally deactivate these two industry-leading models, released less than 72 hours prior, for all non-US users. Anthropic stated that the US government's decision was based on a "serious misunderstanding."

On June 9, Anthropic had announced the launch of its first publicly available "Mythos-class" series model, Claude Fable 5. Simultaneously, the latest version of its frontier model, Claude Mythos 5, would continue to be offered first to a small number of cybersecurity defenders and infrastructure providers through its Project Glasswing program.

According to reports, Fable 5's performance surpasses all previously publicly released models by the company, leading in almost all AI performance benchmarks. Anthropic confidently introduced it, stating: "The longer and more complex the task, the greater Fable 5's lead over other models."

The release of Fable 5 had already caused a stir in the tech community, with industry insiders commenting that Fable 5 already functions like a model capable of working on an engineering site. You throw a task at it, and it can read the project, break down tasks, call tools, fix errors, and keep running. Anthropic's release mentioned that Stripe used Fable 5 for migration within a 50-million-line Ruby codebase, compressing work that would have taken a team two months down to a single day. Even accounting for marketing elements, such cases highlight that AI coding is moving from assisting with writing functions to taking over engineering workflows.

However, the sudden restriction of Anthropic's two flagship models also indicates that the US government's stance towards its most frontier technology is not entirely open. Industry observers believe that against the backdrop of rising uncertainty regarding access to closed-source frontier models overseas, the usability, stability, and value of the developer ecosystem for domestic open-source models like GLM-5.2 are being reassessed.