Traders see an 86.5% chance the U.S. government will not remove public access to a major Chinese AI model such as DeepSeek, Qwen, or Kimi by year-end because recent threats of sanctions and export controls have not produced any verified federal action on downloadable open-weight models. Investigations into data security and IP concerns continue, with targeted restrictions limited to government devices and critical infrastructure rather than broad public bans. Open-source weights prove difficult to enforce against once released, while U.S. companies increasingly integrate these large language models into products, creating practical and economic barriers. Upcoming executive orders or congressional bills could shift sentiment, yet the absence of concrete removals to date anchors the current market-implied odds.