Anthropic's Fable 5 led every major benchmark against GPT-5.5 with 80.3% on SWE-Bench Pro versus 58.6% and topped leaderboards for three days. A US export control order then removed it, leaving GPT-5.5 as the strongest widely available option despite higher costs and narrower capability gaps on some agentic tasks. Enterprises now balance performance against price and policy.