Opus 5 Releases, China Catches Up, and the OpenAI Model Sandbox Escape

<p><a href="https://x.com/theo" target="_blank" rel="ugc noopener noreferrer">Theo</a> and <a href="https://x.com/davis7" target="_blank" rel="ugc noopener noreferrer">Ben</a> break down why Opus 5 feels like GPT-5.5 crossed with Fable rather than GPT-5.6 crossed with Fable, what Fable found when it audited an Opus thread line by line, where Opus still clearly wins (3D, animation, and Claude Code limits at half Fable&#39;s price with no 50% weekly cap), plus Kimi K3, GLM 5.2, the Hugging Face hack, Grok 4.5, Codex, and T3 Code.</p><p>Thanks to this episode&#39;s sponsor, General Translation:</p><ul><li>General Translation: <a href="https://nerdsnipe.link/gt" target="_blank" rel="ugc noopener noreferrer">https://nerdsnipe.link/gt</a></li></ul><p><br></p><p>Sources available on our Substack:</p><p><a href="https://nerdsnipe.substack.com/" target="_blank" rel="noopener noreferer">https://nerdsnipe.substack.com/</a></p>