{"id":23659,"date":"2024-12-23T10:59:36","date_gmt":"2024-12-23T02:59:36","guid":{"rendered":"https:\/\/aif.amtbbs.org\/?p=23659"},"modified":"2024-12-23T10:59:36","modified_gmt":"2024-12-23T02:59:36","slug":"%e5%88%9a%e5%88%9a%ef%bc%8copenai%e6%94%be%e5%87%ba%e6%9c%80%e5%90%8e%e5%a4%a7%e6%83%8a%e5%96%9co3%ef%bc%8c%e9%ab%98%e8%ae%a1%e7%ae%97%e6%a8%a1%e5%bc%8f%e6%af%8f%e4%bb%bb%e5%8a%a1%e8%8a%b1%e8%b4%b9","status":"publish","type":"post","link":"https:\/\/aif.amtbbs.org\/index.php\/2024\/12\/23\/23659\/","title":{"rendered":"\u521a\u521a\uff0cOpenAI\u653e\u51fa\u6700\u540e\u5927\u60ca\u559co3\uff0c\u9ad8\u8ba1\u7b97\u6a21\u5f0f\u6bcf\u4efb\u52a1\u82b1\u8d39\u6570\u5343\u7f8e\u5143"},"content":{"rendered":"<div><img data-dominant-color=\"12558e\" data-has-transparency=\"false\" style=\"--dominant-color: #12558e;\" loading=\"lazy\" decoding=\"async\" class=\"not-transparent alignnone size-full wp-image-23661\" src=\"https:\/\/aiforumimage.oss-cn-shanghai.aliyuncs.com\/wp-content\/uploads\/2024\/12\/e78cfaf7-b238-4bca-b0a8-4798ff9fb2fa-300x167-1.png\" width=\"300\" height=\"167\" alt=\"\" srcset=\"https:\/\/aiforumimage.oss-cn-shanghai.aliyuncs.com\/wp-content\/uploads\/2024\/12\/e78cfaf7-b238-4bca-b0a8-4798ff9fb2fa-300x167-1.png 300w, https:\/\/aiforumimage.oss-cn-shanghai.aliyuncs.com\/wp-content\/uploads\/2024\/12\/e78cfaf7-b238-4bca-b0a8-4798ff9fb2fa-300x167-1-150x84.png 150w\" sizes=\"auto, (max-width: 300px) 100vw, 300px\" \/><\/div>\n<div class=\"article-desc\">o3 \u662f o1 \u7cfb\u5217\u6a21\u578b\u7684\u7ee7\u4efb\u8005\u3002\u8fd9\u7c7b\u6a21\u578b\u7684\u7279\u70b9\u662f\u8ba9\u6a21\u578b\u5728\u56de\u7b54\u95ee\u9898\u4e4b\u524d\u82b1\u66f4\u591a\u65f6\u95f4\u601d\u8003\uff08\u63a8\u7406\uff09\uff0c\u4ece\u800c\u63d0\u9ad8\u56de\u7b54\u7684\u51c6\u786e\u7387\u3002\u4e0d\u8fc7\uff0cOpenAI \u5728\u547d\u540d\u4e0a\u8df3\u8fc7\u4e86 o2\u3002\u636e The Information \u62a5\u9053\uff0c\u8fd9\u4e48\u505a\u662f\u4e3a\u4e86\u907f\u514d\u7248\u6743\u95ee\u9898\uff0c\u56e0\u4e3a\u82f1\u56fd\u6709\u5bb6\u7535\u4fe1\u516c\u53f8\u540d\u53eb O2\uff0c\u53ef\u80fd\u5f15\u8d77\u6df7\u6dc6\u3002Sam Altman \u5728\u4eca\u5929\u4e0b\u5348\u7684\u76f4\u64ad\u4e2d\u8bc1\u5b9e\u4e86\u8fd9\u4e00\u70b9\u3002<\/div>\n<div id=\"postspictures\" class=\"article-content\">\n<div id=\"container\" class=\"container am-engine\" data-v-1d7a5742=\"\" data-element=\"root\">\n<p>\u521a\u521a\uff0cOpenAI \u4e3a\u671f 12 \u5929\u7684\u53d1\u5e03\u8fce\u6765\u5c3e\u58f0\u3002\u5982\u5916\u754c\u6240\u6599\uff0c\u65b0\u7684\u63a8\u7406\u7cfb\u5217\u6a21\u578b \u2014\u2014o3 \u548c o3-mini \u6210\u4e3a\u8fd9\u6b21\u53d1\u5e03\u7684\u6536\u5b98\u4e4b\u4f5c\u3002<\/p>\n<p>o3 \u662f o1 \u7cfb\u5217\u6a21\u578b\u7684\u7ee7\u4efb\u8005\u3002\u8fd9\u7c7b\u6a21\u578b\u7684\u7279\u70b9\u662f\u8ba9\u6a21\u578b\u5728\u56de\u7b54\u95ee\u9898\u4e4b\u524d\u82b1\u66f4\u591a\u65f6\u95f4\u601d\u8003\uff08\u63a8\u7406\uff09\uff0c\u4ece\u800c\u63d0\u9ad8\u56de\u7b54\u7684\u51c6\u786e\u7387\u3002\u4e0d\u8fc7\uff0cOpenAI \u5728\u547d\u540d\u4e0a\u8df3\u8fc7\u4e86 o2\u3002\u636e The Information \u62a5\u9053\uff0c\u8fd9\u4e48\u505a\u662f\u4e3a\u4e86\u907f\u514d\u7248\u6743\u95ee\u9898\uff0c\u56e0\u4e3a\u82f1\u56fd\u6709\u5bb6\u7535\u4fe1\u516c\u53f8\u540d\u53eb O2\uff0c\u53ef\u80fd\u5f15\u8d77\u6df7\u6dc6\u3002Sam Altman \u5728\u4eca\u5929\u4e0b\u5348\u7684\u76f4\u64ad\u4e2d\u8bc1\u5b9e\u4e86\u8fd9\u4e00\u70b9\u3002<\/p>\n<p>\u4e8b\u5b9e\u4e0a\uff0c\u4ece\u6628\u5929\u5f00\u59cb\uff0cOpenAI \u5c31\u5df2\u7ecf\u5f00\u59cb\u9884\u70ed\u8fd9\u4e2a\u6a21\u578b\u3002\u800c\u4e14\u5df2\u7ecf\u6709\u5f00\u53d1\u8005\u5728\u7f51\u4e0a\u627e\u5230\u4e86 OpenAI \u7f51\u7ad9\u4e0a\u5bf9 o3_min_safety_test \u7684\u5f15\u7528\u3002<\/p>\n<p><img decoding=\"async\" title=\"\u56fe\u7247\" src=\"https:\/\/s4.51cto.com\/oss\/202412\/21\/d3c131a79b792630e53496be3413e5d0cb69be.webp\" alt=\"\" data-type=\"block\" \/><\/p>\n<p>\u4e5f\u6709\u4eba\u65e9\u65e9\u5c31\u60f3\u770b\u770b o3 \u7a76\u7adf\u6709\u4f55\u771f\u672c\u4e8b\uff0c\u80fd\u5426\u5339\u654c\u8c37\u6b4c\u6628\u5929\u53d1\u5e03\u7684 Gemini 2.0 Flash Thinking\u3002<\/p>\n<p><img decoding=\"async\" title=\"\u56fe\u7247\" src=\"https:\/\/s7.51cto.com\/oss\/202412\/21\/a6990e66859a7fab2ad611b66004f4ff4fb9f3.webp\" alt=\"\" data-type=\"block\" \/><\/p>\n<p>\u73b0\u5728\uff0c\u548c\u4f20\u8a00\u7684\u4e00\u6837\uff0co3 \u548c o3-mini \u6765\u4e86\uff01\u9057\u61be\u7684\u662f\uff0co3 \u7cfb\u5217\u6a21\u578b\u5e76\u4e0d\u4f1a\u76f4\u63a5\u516c\u5f00\u53d1\u5e03\uff0c\u800c\u662f\u4f1a\u5148\u8fdb\u884c\u5b89\u5168\u6d4b\u8bd5\u3002Sam Altman \u4e5f\u6307\u51fa\u4eca\u5929\u4e0d\u662f\u53d1\u5e03\uff08launch\uff09\uff0c\u53ea\u662f\u5ba3\u5e03\uff08announce\uff09\u3002<\/p>\n<p>Sam Altman \u63d0\u5230\uff0c\u4ed6\u4eec\u8ba1\u5212\u5728\u4e00\u6708\u5e95\u5de6\u53f3\u63a8\u51fa o3-mini\uff0c\u5e76\u5728\u4e0d\u4e45\u540e\u63a8\u51fa\u5b8c\u6574\u7684 o3 \u6a21\u578b\u3002<\/p>\n<h3>o3\uff1a\u7a81\u7834 ARC-AGI \u57fa\u51c6\uff0c\u535a\u58eb\u7ea7\u95ee\u9898\u6c42\u89e3\u80fd\u529b<\/h3>\n<p>\u9996\u5148\uff0c\u5728 12 \u53ea\u5723\u8bde\u9752\u86d9\u805a\u96c6\u7684\u5706\u684c\u524d\uff0cARC Prize Fundation \u603b\u88c1 Greg Kamradt \u53c2\u4e0e\u4e86\u5bf9 o3 \u6a21\u578b\u7684\u4ecb\u7ecd\u3002<\/p>\n<p><img decoding=\"async\" title=\"\u56fe\u7247\" src=\"https:\/\/s5.51cto.com\/oss\/202412\/21\/d27f35d963950849ef084916e008406fe9bb75.webp\" alt=\"\" data-type=\"block\" \/><\/p>\n<p>ARC Prize Fundation \u662f\u4e00\u4e2a\u975e\u8425\u5229\u7ec4\u7ec7\uff0c\u65e8\u5728\u300c\u901a\u8fc7\u57fa\u51c6\u6d4b\u8bd5\u6765\u6210\u4e3a\u5b9e\u73b0 AGI \u4e4b\u8def\u7684\u5317\u6781\u661f\u300d\u3002\u8be5\u7ec4\u7ec7\u7684\u9996\u4e2a\u57fa\u51c6 ARC-AGI \u5df2\u7ecf\u63d0\u51fa\u4e86 5 \u5e74\u65f6\u95f4\uff0c\u4f46\u4e00\u76f4\u672a\u88ab\u653b\u514b\u3002<\/p>\n<p>\u76f4\u5230\u73b0\u5728\uff0cKamradt \u5ba3\u5e03 o3 \u5df2\u7ecf\u5728\u8be5\u57fa\u51c6\u4e0a\u8fbe\u5230\u4e86\u4f18\u826f\u6c34\u5e73\uff0c\u6210\u4e3a\u9996\u4e2a\u7a81\u7834 ARC-AGI \u57fa\u51c6\u7684 AI \u6a21\u578b\u3002<\/p>\n<p><img decoding=\"async\" title=\"\u56fe\u7247\" src=\"https:\/\/s5.51cto.com\/oss\/202412\/21\/016218e97ea66225eb467899a29c722e466fa6.webp\" alt=\"\" data-type=\"block\" \/><\/p>\n<p>\u636e\u4ecb\u7ecd\uff0co3 \u7cfb\u5217\u6a21\u578b\u5728 ARC-AGI \u57fa\u51c6\u4e0a\u7684\u6700\u4f4e\u6027\u80fd\u53ef\u8fbe\u5230 75.7%\uff0c\u800c\u5982\u679c\u8ba9\u5176\u4f7f\u7528\u66f4\u591a\u8ba1\u7b97\u8d44\u6e90\u601d\u8003\u66f4\u957f\u65f6\u95f4\uff0co3 \u66f4\u662f\u53ef\u4ee5\u8fbe\u5230 87.5% \u7684\u6c34\u5e73\u3002<\/p>\n<p>\u5728 ARC-AGI \u57fa\u51c6\u4e2d\uff0cAI \u9700\u8981\u6839\u636e\u914d\u5bf9\u7684\u300c\u8f93\u5165 &#8211; \u8f93\u51fa\u300d\u793a\u4f8b\u5bfb\u627e\u89c4\u5f8b\uff0c\u7136\u540e\u518d\u57fa\u4e8e\u4e00\u4e2a\u8f93\u5165\u9884\u6d4b\u8f93\u51fa\uff0c\u4e0b\u56fe\u5c55\u793a\u4e86\u4e00\u4e9b\u4f8b\u5b50\u3002\u53c2\u52a0\u8fc7\u6bd5\u4e1a\u5b63\u62db\u8058\u6216\u516c\u52a1\u5458\u8003\u8bd5\u7684\u4eba\u6216\u8bb8\u5bf9\u6b64\u7c7b\u56fe\u5f62\u63a8\u7406\u95ee\u9898\u5e76\u4e0d\u964c\u751f\u3002<\/p>\n<p><img decoding=\"async\" title=\"\u56fe\u7247\" src=\"https:\/\/s5.51cto.com\/oss\/202412\/21\/2447a75061c5fc393106981711143a0cf0bd99.webp\" alt=\"\" data-type=\"block\" \/><\/p>\n<p>Greg Kamradt \u4e5f\u987a\u52bf\u5ba3\u5e03\u5f00\u6e90\u4e86\u4e00\u4e2a\u4efb\u52a1\u6570\u636e\u96c6\uff1ahttps:\/\/github.com\/arcprizeorg\/model_baseline<\/p>\n<p><img decoding=\"async\" title=\"\u56fe\u7247\" src=\"https:\/\/s9.51cto.com\/oss\/202412\/21\/2806d1f847d70d363087978b91ad28d3d5b5f3.webp\" alt=\"\" data-type=\"block\" \/><\/p>\n<p>ARC-AGI \u53d1\u8d77\u8005\u3001Keras \u4e4b\u7236 Fran\u00e7ois Chollet \u5728\u6d4b\u8bd5\u62a5\u544a\u4e2d\u5199\u9053\uff0c\u300c\uff08o3 \u7cfb\u5217\u6a21\u578b\uff09\u5728\u9ad8\u6548\u7387\u6a21\u5f0f\u4e0b\u53d6\u5f97\u7684 75.7% \u7684\u5206\u6570\u7b26\u5408 ARC-AGI-Pub \u7684\u9884\u7b97\u89c4\u5219\uff08\u6210\u672c &lt;10000 \u7f8e\u5143\uff09\uff0c\u56e0\u6b64\u6709\u8d44\u683c\u5728\u516c\u5171\u6392\u884c\u699c\u4e0a\u6392\u540d\u7b2c\u4e00\uff01\u800c\u5b83\u4eec\u5728\u4f4e\u6548\u7387\u6a21\u5f0f\u4e0b\u53d6\u5f97\u7684 87.5% \u7684\u5206\u6570\u3002\u867d\u7136\u6210\u672c\u9ad8\u6602\uff0c\u4f46\u4ecd\u7136\u8868\u660e\u65b0\u4efb\u52a1\u7684\u6027\u80fd\u786e\u5b9e\u968f\u7740\u8ba1\u7b97\u91cf\u7684\u589e\u52a0\u800c\u63d0\u9ad8\uff08\u81f3\u5c11\u8fbe\u5230\u4e86\u8fd9\u4e2a\u6c34\u5e73\uff09\u3002\u300d<\/p>\n<p><img decoding=\"async\" title=\"\u56fe\u7247\" src=\"https:\/\/s7.51cto.com\/oss\/202412\/21\/42be85e083deb0af985398f786e2ddff9c6a89.webp\" alt=\"\" data-type=\"block\" \/><\/p>\n<p>\u62a5\u544a\u4e2d\u8fd8\u5199\u9053\uff0c\u5c3d\u7ba1\u6bcf\u9879\u4efb\u52a1\u7684\u6210\u672c\u5f88\u9ad8\uff08o3 \u5728\u4f4e\u8ba1\u7b97\u91cf\u6a21\u5f0f\u4e0b\u6bcf\u4e2a\u4efb\u52a1\u9700\u8981 17-20 \u7f8e\u5143\uff0c\u9ad8\u8ba1\u7b97\u91cf\u6a21\u5f0f\u4e0b\u6bcf\u4e2a\u4efb\u52a1\u6570\u5343\u7f8e\u5143\uff09\uff0c\u4f46\u8fd9\u4e9b\u6570\u5b57\u4e0d\u4ec5\u4ec5\u662f\u5c06\u66b4\u529b\u8ba1\u7b97\u5e94\u7528\u4e8e\u57fa\u51c6\u6d4b\u8bd5\u7684\u7ed3\u679c\u3002OpenAI \u7684\u65b0 o3 \u6a21\u578b\u4ee3\u8868\u4e86\u4eba\u5de5\u667a\u80fd\u9002\u5e94\u65b0\u4efb\u52a1\u7684\u80fd\u529b\u7684\u91cd\u5927\u98de\u8dc3\u3002\u8fd9\u4e0d\u4ec5\u4ec5\u662f\u6e10\u8fdb\u5f0f\u7684\u6539\u8fdb\uff0c\u800c\u662f\u771f\u6b63\u7684\u7a81\u7834\uff0c\u6807\u5fd7\u7740\u4e0e LLM \u4e4b\u524d\u7684\u5c40\u9650\u6027\u76f8\u6bd4\uff0c\u4eba\u5de5\u667a\u80fd\u80fd\u529b\u53d1\u751f\u4e86\u8d28\u7684\u8f6c\u53d8\u3002o3 \u80fd\u591f\u9002\u5e94\u4ee5\u524d\u4ece\u672a\u9047\u5230\u8fc7\u7684\u4efb\u52a1\uff0c\u53ef\u4ee5\u8bf4\u5728 ARC-AGI \u9886\u57df\u63a5\u8fd1\u4eba\u7c7b\u6c34\u5e73\u7684\u8868\u73b0\u3002<\/p>\n<p><img decoding=\"async\" title=\"\u56fe\u7247\" src=\"https:\/\/s3.51cto.com\/oss\/202412\/21\/523eebe947de0ae6f209370e3277c517eca670.webp\" alt=\"\" data-type=\"block\" \/><\/p>\n<p><img decoding=\"async\" title=\"\u56fe\u7247\" src=\"https:\/\/s3.51cto.com\/oss\/202412\/21\/36d2b8d74680c0193727687e4276731efbf116.webp\" alt=\"\" data-type=\"block\" \/><\/p>\n<p>\u4e0d\u8fc7\uff0c\u62a5\u544a\u4e5f\u6307\u51fa\uff0cARC-AGI \u5e76\u4e0d\u662f\u5bf9 AGI \u7684\u4e25\u5cfb\u8003\u9a8c\uff0c\u901a\u8fc7 ARC-AGI \u5e76\u4e0d\u7b49\u4e8e\u5b9e\u73b0 AGI\u3002\u300c\u4e8b\u5b9e\u4e0a\uff0c\u6211\u8ba4\u4e3a o3 \u8fd8\u4e0d\u662f AGI\u3002o3 \u5728\u4e00\u4e9b\u975e\u5e38\u7b80\u5355\u7684\u4efb\u52a1\u4e0a\u4ecd\u7136\u5931\u8d25\uff0c\u8fd9\u8868\u660e\u5176\u4e0e\u4eba\u7c7b\u667a\u80fd\u5b58\u5728\u6839\u672c\u5dee\u5f02\u3002\u300dFran\u00e7ois Chollet \u8868\u793a\u3002<\/p>\n<p><img decoding=\"async\" title=\"\u56fe\u7247\" src=\"https:\/\/s9.51cto.com\/oss\/202412\/21\/c73b5cb3749223061e1084bcd2706bb7c93ee3.webp\" alt=\"\" data-type=\"block\" \/><\/p>\n<p>OpenAI o3 \u6d88\u8017\u6570\u5343\u7f8e\u5143\uff0c\u751f\u6210\u6570\u767e\u4e07\u4e2a token \u4e5f\u6ca1\u80fd\u89e3\u51b3\u7684\u95ee\u9898\u3002<\/p>\n<p>ARC Prize Fundation \u603b\u88c1 Greg Kamradt \u8868\u793a\uff0c\u660e\u5e74\u8fd8\u4f1a\u4e0e OpenAI \u4e00\u8d77\u5f00\u53d1\u4e0b\u4e00\u4ee3\u57fa\u51c6\u3002\u4e0d\u8fc7\uff0c\u4ece\u65e9\u671f\u6570\u636e\u70b9\u6765\u770b\uff0c\u5373\u5c06\u63a8\u51fa\u7684 ARC-AGI-2 \u57fa\u51c6\u6d4b\u8bd5\u4ecd\u5c06\u5bf9 o3 \u6784\u6210\u91cd\u5927\u6311\u6218\uff0c\u5373\u4f7f\u5728\u9ad8\u8ba1\u7b97\u91cf\u4e0b\uff0c\u5176\u5f97\u5206\u4e5f\u53ef\u80fd\u4f1a\u964d\u4f4e\u5230 30% \u4ee5\u4e0b\uff08\u800c\u806a\u660e\u4eba\u5728\u4e0d\u7ecf\u8fc7\u4efb\u4f55\u57f9\u8bad\u7684\u60c5\u51b5\u4e0b\u4ecd\u7136\u80fd\u591f\u5f97\u5206\u8d85\u8fc7 95%\uff09\u3002<\/p>\n<p>\u5728\u62a5\u544a\u4e2d\uff0cFran\u00e7ois Chollet \u8fd8\u5206\u6790\u4e86\u4e3a\u4ec0\u4e48 o3 \u7684\u5206\u6570\u53ef\u4ee5\u63d0\u5347\u90a3\u4e48\u591a\u3002\u4ed6\u5199\u9053\uff0c\u300cLLM \u5c31\u50cf\u4e00\u4e2a\u5411\u91cf\u7a0b\u5e8f\u7684\u5b58\u50a8\u5e93\u3002\u5f53\u88ab\u63d0\u793a\u65f6\uff0c\u5b83\u4eec\u4f1a\u68c0\u7d22\u4f60\u7684\u63d0\u793a\u8bcd\u6620\u5c04\u5230\u7684\u7a0b\u5e8f\uff0c\u5e76\u5728\u5f53\u524d\u7684\u8f93\u5165\u4e0a\u300e\u6267\u884c\u300f\u5b83\u3002LLM \u662f\u4e00\u79cd\u901a\u8fc7\u88ab\u52a8\u63a5\u89e6\u4eba\u7c7b\u751f\u6210\u7684\u5185\u5bb9\u6765\u5b58\u50a8\u548c\u64cd\u4f5c\u5316\u6570\u767e\u4e07\u6709\u7528\u5c0f\u7a0b\u5e8f\u7684\u65b9\u6cd5\u3002\u300d<\/p>\n<p>\u300c\u53ea\u8981\u6709\u8db3\u591f\u7684\u5408\u9002\u8bad\u7ec3\u6570\u636e\uff0c\u8fd9\u79cd\u300e\u8bb0\u5fc6\u3001\u68c0\u7d22\u3001\u5e94\u7528\u300f\u7684\u6a21\u5f0f\u5c31\u80fd\u5728\u4efb\u610f\u4efb\u52a1\u4e0a\u8fbe\u5230\u4efb\u610f\u6c34\u5e73\u7684\u6280\u80fd\uff0c\u4f46\u5b83\u65e0\u6cd5\u9002\u5e94\u65b0\u60c5\u51b5\u6216\u5373\u65f6\u5b66\u4e60\u65b0\u6280\u80fd\uff08\u4e5f\u5c31\u662f\u8bf4\uff0c\u8fd9\u91cc\u5e76\u6ca1\u6709\u6d89\u53ca\u5230 fluid intelligence\uff09\u3002\u8fd9\u4e00\u70b9\u5728 LLM \u5728 ARC-AGI \u4e0a\u7684\u8868\u73b0\u4e2d\u5f97\u5230\u4e86\u4f53\u73b0\uff0cARC-AGI \u662f\u4e13\u95e8\u8bbe\u8ba1\u7528\u6765\u8861\u91cf\u5bf9\u65b0\u4e8b\u7269\u9002\u5e94\u80fd\u529b\u7684\u57fa\u51c6\u6d4b\u8bd5 \u2014\u2014GPT-3 \u5f97\u5206\u4e3a 0\uff0cGPT-4 \u5f97\u5206\u63a5\u8fd1 0\uff0cGPT-4o \u8fbe\u5230\u4e86 5%\u3002\u5c06\u8fd9\u4e9b\u6a21\u578b\u6269\u5c55\u5230\u53ef\u80fd\u7684\u6781\u9650\uff0c\u5e76\u6ca1\u6709\u4f7f ARC-AGI \u7684\u5f97\u5206\u63a5\u8fd1\u51e0\u5e74\u524d\u57fa\u672c\u7684\u66b4\u529b\u679a\u4e3e\u5c31\u80fd\u8fbe\u5230\u7684\u6c34\u5e73\uff08\u9ad8\u8fbe 50%\uff09\u3002\u300d<\/p>\n<p>\u300c\u8981\u9002\u5e94\u65b0\u4e8b\u7269\uff0c\u4f60\u9700\u8981\u4e24\u6837\u4e1c\u897f\u3002\u9996\u5148\uff0c\u4f60\u9700\u8981\u77e5\u8bc6 \u2014\u2014 \u4e00\u5957\u53ef\u590d\u7528\u7684\u51fd\u6570\u6216\u7a0b\u5e8f\u5e93\u3002LLM \u5728\u8fd9\u65b9\u9762\u62e5\u6709\u7684\u8d44\u6e90\u7ef0\u7ef0\u6709\u4f59\u3002\u5176\u6b21\uff0c\u4f60\u9700\u8981\u5728\u9762\u5bf9\u65b0\u4efb\u52a1\u65f6\uff0c\u80fd\u591f\u5c06\u8fd9\u4e9b\u51fd\u6570\u91cd\u65b0\u7ec4\u5408\u6210\u4e00\u4e2a\u5168\u65b0\u7684\u7a0b\u5e8f \u2014\u2014 \u4e00\u4e2a\u80fd\u591f\u6a21\u62df\u5f53\u524d\u4efb\u52a1\u7684\u7a0b\u5e8f\u3002\u8fd9\u5c31\u662f\u7a0b\u5e8f\u5408\u6210\u3002LLM \u957f\u671f\u4ee5\u6765\u7f3a\u4e4f\u8fd9\u4e00\u7279\u6027\u3002\u800c o \u7cfb\u5217\u6a21\u578b\u89e3\u51b3\u4e86\u8fd9\u4e2a\u95ee\u9898\u3002\u300d<\/p>\n<p>\u300c\u5173\u4e8e o3 \u6a21\u578b\u7684\u5177\u4f53\u5de5\u4f5c\u539f\u7406\uff0c\u76ee\u524d\u6211\u4eec\u53ea\u80fd\u8fdb\u884c\u4e00\u4e9b\u63a8\u6d4b\u3002o3 \u6a21\u578b\u7684\u6838\u5fc3\u673a\u5236\u4f3c\u4e4e\u662f\u5728 token \u7a7a\u95f4\u5185\u8fdb\u884c\u81ea\u7136\u8bed\u8a00\u7a0b\u5e8f\u641c\u7d22\u548c\u6267\u884c \u2014\u2014 \u5728\u6d4b\u8bd5\u65f6\uff0c\u6a21\u578b\u4f1a\u5728\u53ef\u80fd\u7684\u601d\u7ef4\u94fe\u7a7a\u95f4\u4e2d\u641c\u7d22\uff0c\u8fd9\u4e9b\u601d\u7ef4\u94fe\u63cf\u8ff0\u4e86\u89e3\u51b3\u4efb\u52a1\u6240\u9700\u7684\u6b65\u9aa4\uff0c\u8fd9\u79cd\u65b9\u5f0f\u53ef\u80fd\u4e0e AlphaZero \u98ce\u683c\u7684\u8499\u7279\u5361\u6d1b\u6811\u641c\u7d22\uff08Monte-Carlo tree search\uff09\u9887\u6709\u76f8\u4f3c\u4e4b\u5904\u3002\u5728 o3 \u7684\u60c5\u51b5\u4e0b\uff0c\u641c\u7d22\u53ef\u80fd\u7531\u67d0\u79cd\u8bc4\u4f30\u6a21\u578b\u5f15\u5bfc\u8fdb\u884c\u3002\u503c\u5f97\u6ce8\u610f\u7684\u662f\uff0cDeepMind \u7684 Demis Hassabis \u5728 2023 \u5e74 6 \u6708\u7684\u4e00\u6b21\u91c7\u8bbf\u4e2d\u6697\u793a\uff0cDeepMind \u4e00\u76f4\u5728\u7814\u7a76\u8fd9\u4e00\u6982\u5ff5 \u2014\u2014 \u8fd9\u9879\u5de5\u4f5c\u5df2\u7ecf\u915d\u917f\u4e86\u5f88\u957f\u65f6\u95f4\u3002\u300d<\/p>\n<p><img decoding=\"async\" title=\"\u56fe\u7247\" src=\"https:\/\/s2.51cto.com\/oss\/202412\/21\/67b16bd87087deae6948447764d84fb66f6959.webp\" alt=\"\" data-type=\"block\" \/><\/p>\n<p><img decoding=\"async\" title=\"\u56fe\u7247\" src=\"https:\/\/s3.51cto.com\/oss\/202412\/21\/55e9d3f99a47fa3c29b583ffa68d8a7b56a64b.webp\" alt=\"\" data-type=\"block\" \/><\/p>\n<p>\u8be6\u7ec6\u5206\u6790\u53c2\u89c1\uff1ahttps:\/\/arcprize.org\/blog\/oai-o3-pub-breakthrough<\/p>\n<p>\u9664\u4e86\u89e3\u51b3 ARC-AGI \u8fd9\u6837\u7684\u300cIQ\u300d \u95ee\u9898\uff0co3 \u7684\u7f16\u7801\u80fd\u529b\u4e5f\u6bd4\u4e4b\u524d\u7684 o1 \u7cfb\u5217\u660e\u663e\u66f4\u80dc\u4e00\u7b79\u3002\u5728 SWE-bench Verified \u57fa\u51c6\u4e0a\uff0co3 \u7684\u51c6\u786e\u7387\u7ea6\u4e3a 71.7%\uff0c\u6bd4 o1 \u6a21\u578b\u9ad8\u51fa 20% \u4ee5\u4e0a\u3002\u5728 Competition Code \u4e2d\uff0co3 \u83b7\u5f97\u4e86\u96be\u4ee5\u7f6e\u4fe1 2727 Elo \u5f97\u5206\uff0c\u800c o1 \u4ec5\u4e3a 1891\u3002<\/p>\n<p><img decoding=\"async\" title=\"\u56fe\u7247\" src=\"https:\/\/s7.51cto.com\/oss\/202412\/21\/c7f317b3179dd474c31597cfe8aabe2ec7ea7e.webp\" alt=\"\" data-type=\"block\" \/><\/p>\n<p>\u6b64\u5916\uff0co3 \u5728\u7ade\u8d5b\u6570\u5b66\uff08Competition Math \uff09\u4e0a\u7684\u51c6\u786e\u7387\u8fbe\u5230 96.7%\uff0c\u5728 GPQA Diamond \uff08\u8be5\u57fa\u51c6\u8861\u91cf\u6a21\u578b\u5728\u535a\u58eb\u7ea7\u79d1\u5b66\u95ee\u9898\u4e0a\u7684\u8868\u73b0\uff09\u4e0a\u7684\u51c6\u786e\u7387\u8fbe\u5230 87.7%\uff0c\u6bd4\u4e4b\u524d\u7684 o1 \u8868\u73b0\uff0878%\uff09\u9ad8\u51fa\u8fd1 10%\u3002\u603b\u7ed3\u6765\u770b\uff0co3 \u5728\u8fd9\u4e24\u9879\u57fa\u51c6\u6d4b\u8bd5\u4e0a\u90fd\u8fbe\u5230\u4e86\u6700\u4f73\u8868\u73b0\u3002<\/p>\n<p><img decoding=\"async\" title=\"\u56fe\u7247\" src=\"https:\/\/s5.51cto.com\/oss\/202412\/21\/b1049087889a79efeff8379fac58b91bac2cd8.webp\" alt=\"\" data-type=\"block\" \/><\/p>\n<p>o3 \u5728 OpenAI \u7814\u7a76\u4eba\u5458\u8ba4\u4e3a\u6700\u4e25\u683c\u7684\u57fa\u51c6\u4e4b\u4e00\uff08EpochAI Frontier Math\uff09\u4e0a\u7684\u8868\u73b0\u5982\u4e0b\u8868\u6240\u793a\u3002\u8be5\u6570\u636e\u96c6\u7531\u65b0\u9896\u7684\u3001\u672a\u53d1\u8868\u7684\u4ee5\u53ca\u975e\u5e38\u96be\u3001\u6781\u5176\u96be\u7684\u95ee\u9898\u7ec4\u6210\u3002<\/p>\n<p>OpenAI \u9ad8\u7ea7\u7814\u7a76\u526f\u603b\u88c1 Mark Chen \u7279\u522b\u5f3a\u8c03\u4e86\uff0c\u5373\u4f7f\u662f\u4e13\u4e1a\u6570\u5b66\u5bb6\u4e5f\u9700\u8981\u51e0\u4e2a\u5c0f\u65f6\u751a\u81f3\u51e0\u5929\u624d\u80fd\u89e3\u51b3\u91cc\u9762\u7684\u95ee\u9898\u4e4b\u4e00\u3002\u800c\u4eca\u5929\uff0c\u6240\u6709\u4ea7\u54c1\u5728\u8fd9\u4e2a\u57fa\u51c6\u4e0a\u7684\u51c6\u786e\u7387\u90fd\u4f4e\u4e8e 2%\uff0co3 \u7684\u51c6\u786e\u7387\u53ef\u4ee5\u8d85\u8fc7 25%\u3002<\/p>\n<p><img decoding=\"async\" title=\"\u56fe\u7247\" src=\"https:\/\/s2.51cto.com\/oss\/202412\/21\/4959b2896ab8c18b741649f1672b6e573c54b9.webp\" alt=\"\" data-type=\"block\" \/><\/p>\n<h3>o3-mini\uff1a\u9ad8\u6027\u80fd\u4f4e\u6210\u672c\uff0c\u975e\u5e38\u9002\u5408\u7f16\u7a0b<\/h3>\n<p>o3 mini \u662f\u4e00\u4e2a\u66f4\u7ecf\u6d4e\u9ad8\u6548\u7684 o3 \u7248\u672c\uff0c\u4e13\u6ce8\u5728\u63d0\u5347\u63a8\u7406\u901f\u5ea6\u3001\u964d\u4f4e\u63a8\u7406\u6210\u672c\u7684\u540c\u65f6\u517c\u987e\u6a21\u578b\u6027\u80fd\u3002<\/p>\n<p><img decoding=\"async\" title=\"\u56fe\u7247\" src=\"https:\/\/s4.51cto.com\/oss\/202412\/21\/845bd383829d53bfcba0427179c5b6bead4750.webp\" alt=\"\" data-type=\"block\" \/><\/p>\n<p>\u4ecb\u7ecd o3-mini \u7684\u662f\u53c2\u4e0e\u8bad\u7ec3\u7684 OpenAI \u7814\u7a76\u8005 Hongyu Ren<\/p>\n<p>\u5b83\u652f\u6301\u4e09\u79cd\u4e0d\u540c\u7684\u63a8\u7406\u65f6\u95f4\u9009\u9879 \u2014\u2014 \u4f4e\u3001\u4e2d\u3001\u9ad8\u3002<\/p>\n<p>\u4e0e o1 \u76f8\u6bd4\uff0co3-mini \u5728 Codeforces \u4e0a\u7684\u6027\u80fd\u5177\u6709\u663e\u8457\u7684\u6210\u672c\u6548\u76ca\uff0c\u8fd9\u4f7f\u5176\u6210\u4e3a\u4e00\u4e2a\u975e\u5e38\u9002\u5408\u7528\u6765\u7f16\u7a0b\u7684\u6a21\u578b\u3002<\/p>\n<p><img decoding=\"async\" title=\"\u56fe\u7247\" src=\"https:\/\/s6.51cto.com\/oss\/202412\/21\/c1aa69950017909dd4a007fb90851c6b9fc9d4.webp\" alt=\"\" data-type=\"block\" \/><\/p>\n<p>\u5728\u6570\u5b66\u95ee\u9898\u4e0a\uff0co3-mini (low) \u5b9e\u73b0\u4e86\u4e0e gpt-4o \u76f8\u5f53\u7684\u4f4e\u5ef6\u8fdf\u3002<\/p>\n<p><img decoding=\"async\" title=\"\u56fe\u7247\" src=\"https:\/\/s6.51cto.com\/oss\/202412\/21\/724ce9b2927e5beb72015582b9447356e0d302.webp\" alt=\"\" data-type=\"block\" \/><\/p>\n<p>o3-mini \u4e0a\u6240\u6709\u7684 API \u7279\u6027\u4ee5\u53ca\u76f8\u5e94\u7684\u80fd\u529b\u8868\u73b0\u5982\u4e0b\uff1a<\/p>\n<p><img decoding=\"async\" title=\"\u56fe\u7247\" src=\"https:\/\/s6.51cto.com\/oss\/202412\/21\/14cc06e77dc07e2de50143a27e7eebb80fde7b.webp\" alt=\"\" data-type=\"block\" \/><\/p>\n<p>Hongyu Ren \u73b0\u573a\u6f14\u793a\u4e86\u51e0\u4e2a\u793a\u4f8b\u3002\u9996\u5148\uff0c\u9488\u5bf9 o3-mini (high)\uff0c\u4efb\u52a1\u662f\u4f7f\u7528 Python \u8bed\u8a00\u7f16\u5199\u4e00\u4e2a\u672c\u5730\u670d\u52a1\u5668\uff0c\u5176\u6709\u4e00\u4e2a\u7b80\u6613\u7684\u7528\u6237 UI\uff0c\u540c\u65f6\u53ef\u4ee5\u5c06\u7528\u6237\u7ed9\u51fa\u7684\u6587\u672c\u6d88\u606f\u901a\u8fc7 API \u4f20\u8f93\u7ed9 o3-mini \u7684 medium \u9009\u9879\uff0c\u7136\u540e\u83b7\u53d6\u5f97\u5230\u7684\u4ee3\u7801\uff0c\u5c06\u5176\u4fdd\u5b58\u5230\u684c\u9762\u7684\u4e00\u4e2a\u4e34\u65f6\u6587\u4ef6\u4e2d\uff0c\u7136\u540e\u5728\u4e00\u4e2a\u65b0\u7684 Python \u7ec8\u7aef\u4e2d\u6267\u884c\u8be5\u6587\u4ef6\u3002<\/p>\n<p><img decoding=\"async\" title=\"\u56fe\u7247\" src=\"https:\/\/s9.51cto.com\/oss\/202412\/21\/f1bbbc9342576ad61889837a150c3d5ac5fb73.webp\" alt=\"\" data-type=\"block\" \/><\/p>\n<p>38 \u79d2\u540e\uff0co3-mini \u751f\u6210\u4e86\u7ed3\u679c\u4ee3\u7801\uff1a<\/p>\n<p><img decoding=\"async\" title=\"\u56fe\u7247\" src=\"https:\/\/s2.51cto.com\/oss\/202412\/21\/61096bb600f88c94380633984de9993f87005f.webp\" alt=\"\" data-type=\"block\" \/><\/p>\n<p>\u63a5\u7740\uff0cHongyu Ren \u76f4\u63a5\u5c06\u5176\u590d\u5236\u5230\u4e86\u4e00\u4e2a server.py \u6587\u4ef6\u4e2d\uff0c\u7136\u540e\u76f4\u63a5\u8fd0\u884c\u4e86\u5b83\u3002<\/p>\n<p><img data-dominant-color=\"dddddd\" data-has-transparency=\"false\" style=\"--dominant-color: #dddddd;\" loading=\"lazy\" decoding=\"async\" class=\"not-transparent alignnone size-full wp-image-23662\" src=\"https:\/\/aiforumimage.oss-cn-shanghai.aliyuncs.com\/wp-content\/uploads\/2024\/12\/647c6fb33f6408c39f6421fff26c8745d5731e.gif\" width=\"1080\" height=\"563\" alt=\"\u56fe\u7247\" title=\"\u56fe\u7247\" \/><\/p>\n<p>\u4e4b\u540e\uff0c\u4ed6\u5c1d\u8bd5\u4e86\u4f7f\u7528\u8fd9\u4e2a\u670d\u52a1\u5668\uff1aprint openai \u5e76\u968f\u673a\u7ed9\u51fa\u4e00\u4e2a\u6570\uff0c\u7b80\u5355\u4efb\u52a1\uff0c\u5f53\u7136\u5f88\u6210\u529f\u3002<\/p>\n<p><img data-dominant-color=\"d9d9d9\" data-has-transparency=\"false\" style=\"--dominant-color: #d9d9d9;\" loading=\"lazy\" decoding=\"async\" class=\"not-transparent alignnone size-full wp-image-23663\" src=\"https:\/\/aiforumimage.oss-cn-shanghai.aliyuncs.com\/wp-content\/uploads\/2024\/12\/573ed4e14470c8582f1919ef637267dbca9ed4.gif\" width=\"1079\" height=\"562\" alt=\"\u56fe\u7247\" title=\"\u56fe\u7247\" \/><\/p>\n<p>\u63a5\u4e0b\u6765\u4ed6\u4eec\u4e0a\u4e86\u70b9\u96be\u5ea6\uff1a\u5728\u76f8\u5f53\u56f0\u96be\u7684 GPQA Diamond \u6570\u636e\u96c6\uff08\u8fd9\u662f\u4e00\u4e2a\u535a\u58eb\u6c34\u5e73\u7684\u79d1\u5b66\u95ee\u9898\u57fa\u51c6\uff09\u4e0a\uff0c\u751f\u6210\u4e00\u4e2a\u4ee3\u7801\u6765\u8bc4\u4f30 o3-mini \u7684 low \u6a21\u5f0f\uff0c\u9700\u8981\u4e0b\u8f7d https:\/\/openaipublic.blob.core.windows.net\/simple-evals\/gpga_diamond.csv<\/p>\n<p>\u6700\u540e\uff0c\u4f7f\u7528\u524d\u9762\u7f16\u5199\u7684 UI \u754c\u9762\uff0co3-mini \u7528\u4e86\u4e00\u5b9a\u65f6\u95f4\u5b8c\u6210\u4e86\u8fd9\u4e2a\u4efb\u52a1\uff0c\u6700\u7ec8\u5f97\u5230 low \u6a21\u5f0f\u4e0b\u7684 o3-mini \u5728 GPQA Diamond \u6570\u636e\u96c6\u4e0a\u7684\u51c6\u786e\u5ea6\u4e3a 61.62%\u3002\u4e5f\u5c31\u662f\u8bf4\uff0c\u4ed6\u4eec\u6210\u529f\u8ba9\u6a21\u578b\u7f16\u5199\u4e86\u4e00\u4e2a\u6d4b\u8bd5\u8be5\u6a21\u578b\u81ea\u8eab\u7684\u811a\u672c\u3002<\/p>\n<p><img data-dominant-color=\"d2d2d2\" data-has-transparency=\"false\" style=\"--dominant-color: #d2d2d2;\" loading=\"lazy\" decoding=\"async\" class=\"not-transparent alignnone size-full wp-image-23664\" src=\"https:\/\/aiforumimage.oss-cn-shanghai.aliyuncs.com\/wp-content\/uploads\/2024\/12\/05549b75522566867bb0469d27eb143bd15696.gif\" width=\"1079\" height=\"561\" alt=\"\u56fe\u7247\" title=\"\u56fe\u7247\" \/><\/p>\n<p>Mark Chen \u6253\u8da3\u5730\u8868\u793a\u660e\u5e74\u8981\u8ba9 AI \u6f14\u793a\u4e00\u4e0b\u81ea\u6211\u63d0\u5347\uff0c\u4f46\u8fd9\u4e2a\u8bf4\u6cd5\u4f3c\u4e4e\u4e0d\u591f PR\uff0c\u8ba9 Sam Altman \u8d76\u5fd9\u8bf4\u4e86\u4e00\u53e5\uff1a\u300cMaybe not.\u300d<\/p>\n<p>\u53e6\u5916\uff0c\u4ed6\u4eec\u4e5f\u7ed9\u51fa\u4e86\u5728 GPQA Diamond \u6570\u636e\u96c6\u4e0a\u7684\u66f4\u591a\u6d4b\u8bd5\u7ed3\u679c\u3002\u53ef\u4ee5\u770b\u5230\uff0co3-mini (low) \u7684\u7ed3\u679c\u4e0e\u4e0a\u9762\u6f14\u793a\u793a\u4f8b\u5f97\u5230\u7684\u7ed3\u679c\u57fa\u672c\u4e00\u81f4\uff0c\u800c o3 \u6ee1\u8840\u7248\u66f4\u662f\u53ef\u4ee5\u5f97\u5230 87.7 \u7684\u9ad8\u51c6\u786e\u5ea6\u5206\u6570\u3002<\/p>\n<p><img decoding=\"async\" title=\"\u56fe\u7247\" src=\"https:\/\/s5.51cto.com\/oss\/202412\/21\/72daeb474fb9156e52640809fcdff207189ba2.webp\" alt=\"\" data-type=\"block\" \/><\/p>\n<h3>\u5982\u4f55\u7533\u8bf7\u6d4b\u8bd5 o3 \u548c o3-mini<\/h3>\n<p>Mark Chen \u8868\u793a\uff0cOpenAI \u5df2\u7ecf\u505a\u4e86\u5927\u91cf\u5185\u90e8\u5b89\u5168\u6d4b\u8bd5\uff0c\u800c\u73b0\u5728\u4ed6\u4eec\u6b63\u5728\u63a8\u8fdb\u5916\u90e8\u5b89\u5168\u6d4b\u8bd5\u3002\u4ece\u5373\u65e5\u8d77\uff0c\u60f3\u8981\u53c2\u4e0e\u7684\u6d4b\u8bd5\u8005\u53ef\u4ee5\u7533\u8bf7\u6d4b\u8bd5 o3-mini\uff0c\u81f3\u4e8e o3\uff0c\u5c31\u8fde\u6d4b\u8bd5\u5458\u4e5f\u8fd8\u5f97\u7ee7\u7eed\u7b49\u5f85\u3002<\/p>\n<p>\u65e9\u671f\u8bbf\u95ee\u7533\u8bf7\u73b0\u5df2\u5728 OpenAI \u7f51\u7ad9\u4e0a\u5f00\u653e\uff0c\u5e76\u5c06\u4e8e 2025 \u5e74 1 \u6708 10 \u65e5\u5173\u95ed\u3002<\/p>\n<p>\u7533\u8bf7\u5730\u5740\uff1ahttps:\/\/openai.com\/index\/early-access-for-safety-testing\/<\/p>\n<p>\u7533\u8bf7\u8005\u5fc5\u987b\u586b\u5199\u4e00\u4efd\u5728\u7ebf\u8868\u683c\uff0c\u8868\u683c\u4e2d\u4f1a\u8981\u6c42\u4ed6\u4eec\u63d0\u4f9b\u5404\u79cd\u4e0d\u540c\u7684\u4fe1\u606f\uff0c\u5305\u62ec\u4e4b\u524d\u53d1\u8868\u7684\u8bba\u6587\u94fe\u63a5\u53ca\u5176\u5728 Github \u4e0a\u7684\u4ee3\u7801\u5e93\uff0c\u5e76\u9009\u62e9\u4ed6\u4eec\u5e0c\u671b\u6d4b\u8bd5\u7684\u6a21\u578b\uff08o3 \u6216 o3-mini\uff09\u4ee5\u53ca\u8ba1\u5212\u4f7f\u7528\u5b83\u4eec\u505a\u4ec0\u4e48\u3002<\/p>\n<p>\u9009\u5b9a\u7684\u7814\u7a76\u4eba\u5458\u5c06\u88ab\u6388\u4e88\u8bbf\u95ee o3 \u548c o3-mini \u7684\u6743\u9650\uff0c\u4ee5\u63a2\u7d22\u5b83\u4eec\u7684\u80fd\u529b\u5e76\u4e3a\u5b89\u5168\u8bc4\u4f30\u505a\u51fa\u8d21\u732e\uff0c\u4e0d\u8fc7 OpenAI \u7684\u8868\u683c\u63d0\u793a\u79f0\uff0co3 \u5c06\u5728\u51e0\u5468\u5185\u65e0\u6cd5\u4f7f\u7528\u3002<\/p>\n<p><img decoding=\"async\" title=\"\u56fe\u7247\" src=\"https:\/\/s4.51cto.com\/oss\/202412\/21\/9942eba083e47ef5570886d5307e5b68a89964.webp\" alt=\"\" data-type=\"block\" \/><\/p>\n<p>OpenAI \u8868\u793a\u4ed6\u4eec\u5c06\u6eda\u52a8\u5ba1\u6838\u7533\u8bf7\uff0c\u5e76\u7acb\u5373\u5f00\u59cb\u9009\u62d4\u7533\u8bf7\u4eba\u3002<\/p>\n<p>\u6700\u540e\uff0cMark Chen \u4e5f\u7b80\u5355\u4ecb\u7ecd\u4e86\u4ed6\u4eec\u7684\u4e00\u79cd\u65b0\u7684\u5b89\u5168\u8bc4\u4f30\u65b9\u6cd5\uff1adeliberative alignment\uff0c\u5373\u5ba1\u8bae\u5f0f\u5bf9\u9f50\u3002\u8fd9\u662f\u4e00\u79cd\u76f4\u63a5\u6559\u6388\u6a21\u578b\u5b89\u5168\u89c4\u8303\u7684\u65b0\u8303\u5f0f\uff0c\u5e76\u53ef\u8bad\u7ec3\u6a21\u578b\u5728\u56de\u7b54\u4e4b\u524d\u660e\u786e\u56de\u5fc6\u89c4\u8303\u5e76\u51c6\u786e\u5730\u6267\u884c\u63a8\u7406\u3002\u4ed6\u4eec\u4f7f\u7528\u4e86\u8fd9\u79cd\u65b9\u6cd5\u6765\u5bf9\u9f50 OpenAI \u7684 o \u7cfb\u5217\u6a21\u578b \uff0c\u5e76\u5b9e\u73b0\u4e86\u5bf9 OpenAI \u5b89\u5168\u653f\u7b56\u7684\u9ad8\u5ea6\u7cbe\u786e\u9075\u5b88\uff0c\u5e76\u4e14\u8fd9\u4e2a\u8fc7\u7a0b\u65e0\u9700\u4eba\u5de5\u7f16\u5199\u7684\u601d\u8def\u6216\u7b54\u6848\u3002<\/p>\n<p><img decoding=\"async\" title=\"\u56fe\u7247\" src=\"https:\/\/s6.51cto.com\/oss\/202412\/21\/1274877913dd247bbc9999f6b25576dd7f0085.webp\" alt=\"\" data-type=\"block\" \/><\/p>\n<p>\u611f\u5174\u8da3\u7684\u8bfb\u8005\u53ef\u67e5\u770b OpenAI \u7684\u76f8\u5173\u8bba\u6587\u3002<\/p>\n<p><img decoding=\"async\" title=\"\u56fe\u7247\" src=\"https:\/\/s9.51cto.com\/oss\/202412\/21\/d30f10e060b153b62770095462e859d88ca5d1.webp\" alt=\"\" data-type=\"block\" \/><\/p>\n<ul data-id=\"ua73dd4b-65nlcF27\">\n<li data-id=\"ld70c578-pgAzNlqu\">\u8bba\u6587\u6807\u9898\uff1aDeliberative Alignment: Reasoning Enables Safer Language Models<\/li>\n<li data-id=\"ld70c578-iDgaXDhh\">\u8bba\u6587\u5730\u5740\uff1ahttps:\/\/assets.ctfassets.net\/kftzwdyauwt9\/4pNYAZteAQXWtloDdANQ7L\/978a6fd0a2ee268b2cb59637bd074cca\/OpenAI_Deliberative-Alignment-Reasoning-Enables-Safer_Language-Models_122024.pdf<\/li>\n<\/ul>\n<p>\u6240\u4ee5\uff0c\u603b\u7ed3\u8d77\u6765\uff0c12 \u5929\u53d1\u5e03\u7684\u6700\u540e\u4e00\u5929\uff0cOpenAI \u771f\u6b63\u53d1\u5e03\u51fa\u6765\u7684\u4e1c\u897f\u53ea\u6709\u4e00\u7bc7\u8bba\u6587\uff0c\u66f4\u591a\u662f\u5151\u73b0\u671f\u7968\u3002\u53e6\u5916\uff0c\u6211\u4eec\u8fd8\u662f\u4e0d\u77e5\u9053 o3 \u662f\u5426\u80fd\u89e3\u7b54\u6700\u96be\u7684\u9ad8\u8003\u6570\u5b66\u9898\uff0c\u4f46\u81f3\u5c11\u4ece OpenAI \u7684\u63cf\u8ff0\u6765\u770b\uff0c\u7b54\u6848\u5e94\u8be5\u662f\u4e50\u89c2\u7684\u3002<\/p>\n<p>\u5bf9\u4e8e OpenAI \u7684\u8fd9\u6700\u540e\u4e00\u5929\u53d1\u5e03\u548c o3 \u6a21\u578b\uff0c\u4f60\u6709\u4ec0\u4e48\u770b\u6cd5\uff1f<\/p>\n<p>\u6587\u7ae0\u6765\u81ea\uff1a51CTO<\/p>\n<\/div>\n<\/div>\n<div class=\"pvc_clear\"><\/div>\n<p id=\"pvc_stats_23659\" class=\"pvc_stats total_only  \" data-element-id=\"23659\" style=\"\"><i class=\"pvc-stats-icon medium\" aria-hidden=\"true\"><svg xmlns=\"http:\/\/www.w3.org\/2000\/svg\" version=\"1.0\" viewBox=\"0 0 502 315\" preserveAspectRatio=\"xMidYMid meet\"><g transform=\"translate(0,332) scale(0.1,-0.1)\" fill=\"\" stroke=\"none\"><path d=\"M2394 3279 l-29 -30 -3 -207 c-2 -182 0 -211 15 -242 39 -76 157 -76 196 0 15 31 17 60 15 243 l-3 209 -33 29 c-26 23 -41 29 -80 29 -41 0 -53 -5 -78 -31z\"\/><path d=\"M3085 3251 c-45 -19 -58 -50 -96 -229 -47 -217 -49 -260 -13 -295 52 -53 146 -42 177 20 16 31 87 366 87 410 0 70 -86 122 -155 94z\"\/><path d=\"M1751 3234 c-13 -9 -29 -31 -37 -50 -12 -29 -10 -49 21 -204 19 -94 39 -189 45 -210 14 -50 54 -80 110 -80 34 0 48 6 76 34 21 21 34 44 34 59 0 14 -18 113 -40 219 -37 178 -43 195 -70 221 -36 32 -101 37 -139 11z\"\/><path d=\"M1163 3073 c-36 -7 -73 -59 -73 -102 0 -56 133 -378 171 -413 34 -32 83 -37 129 -13 70 36 67 87 -16 290 -86 209 -89 214 -129 231 -35 14 -42 15 -82 7z\"\/><path d=\"M3689 3066 c-15 -9 -33 -30 -42 -48 -48 -103 -147 -355 -147 -375 0 -98 131 -148 192 -74 13 15 57 108 97 206 80 196 84 226 37 273 -30 30 -99 39 -137 18z\"\/><path d=\"M583 2784 c-38 -19 -67 -74 -58 -113 9 -42 211 -354 242 -373 16 -10 45 -18 66 -18 51 0 107 52 107 100 0 39 -1 41 -124 234 -80 126 -108 162 -133 173 -41 17 -61 16 -100 -3z\"\/><path d=\"M4250 2784 c-14 -9 -74 -91 -133 -183 -95 -150 -107 -173 -107 -213 0 -55 33 -94 87 -104 67 -13 90 8 211 198 130 202 137 225 78 284 -27 27 -42 34 -72 34 -22 0 -50 -8 -64 -16z\"\/><path d=\"M2275 2693 c-553 -48 -1095 -270 -1585 -649 -135 -104 -459 -423 -483 -476 -23 -49 -22 -139 2 -186 73 -142 361 -457 571 -626 285 -228 642 -407 990 -497 242 -63 336 -73 660 -74 310 0 370 5 595 52 535 111 1045 392 1455 803 122 121 250 273 275 326 19 41 19 137 0 174 -41 79 -309 363 -465 492 -447 370 -946 591 -1479 653 -113 14 -422 18 -536 8z m395 -428 c171 -34 330 -124 456 -258 112 -119 167 -219 211 -378 27 -96 24 -300 -5 -401 -72 -255 -236 -447 -474 -557 -132 -62 -201 -76 -368 -76 -167 0 -236 14 -368 76 -213 98 -373 271 -451 485 -162 444 86 934 547 1084 153 49 292 57 452 25z m909 -232 c222 -123 408 -262 593 -441 76 -74 138 -139 138 -144 0 -16 -233 -242 -330 -319 -155 -123 -309 -223 -461 -299 l-81 -41 32 46 c18 26 49 83 70 128 143 306 141 649 -6 957 -25 52 -61 116 -79 142 l-34 47 45 -20 c26 -10 76 -36 113 -56z m-2057 25 c-40 -58 -105 -190 -130 -263 -110 -324 -59 -707 132 -981 25 -35 42 -64 37 -64 -19 0 -241 119 -326 174 -188 122 -406 314 -532 468 l-58 71 108 103 c185 178 428 349 672 473 66 33 121 60 123 61 2 0 -10 -19 -26 -42z\"\/><path d=\"M2375 1950 c-198 -44 -350 -190 -395 -379 -18 -76 -8 -221 19 -290 114 -284 457 -406 731 -260 98 52 188 154 231 260 27 69 37 214 19 290 -38 163 -166 304 -326 360 -67 23 -215 33 -279 19z\"\/><\/g><\/svg><\/i> <img loading=\"lazy\" decoding=\"async\" width=\"16\" height=\"16\" alt=\"Loading\" src=\"https:\/\/aif.amtbbs.org\/wp-content\/plugins\/page-views-count\/ajax-loader-2x.gif\" border=0 \/><\/p>\n<div class=\"pvc_clear\"><\/div>\n","protected":false},"excerpt":{"rendered":"<p>o3 \u662f o1 \u7cfb\u5217\u6a21\u578b\u7684\u7ee7\u4efb\u8005\u3002\u8fd9\u7c7b\u6a21\u578b\u7684\u7279\u70b9\u662f\u8ba9\u6a21\u578b\u5728\u56de\u7b54\u95ee\u9898\u4e4b\u524d\u82b1\u66f4\u591a\u65f6\u95f4\u601d\u8003\uff08\u63a8\u7406\uff09\uff0c\u4ece\u800c\u63d0\u9ad8\u56de\u7b54\u7684\u51c6 [&hellip;]<\/p>\n<div class=\"pvc_clear\"><\/div>\n<p id=\"pvc_stats_23659\" class=\"pvc_stats total_only  \" data-element-id=\"23659\" style=\"\"><i class=\"pvc-stats-icon medium\" aria-hidden=\"true\"><svg xmlns=\"http:\/\/www.w3.org\/2000\/svg\" version=\"1.0\" viewBox=\"0 0 502 315\" preserveAspectRatio=\"xMidYMid meet\"><g transform=\"translate(0,332) scale(0.1,-0.1)\" fill=\"\" stroke=\"none\"><path d=\"M2394 3279 l-29 -30 -3 -207 c-2 -182 0 -211 15 -242 39 -76 157 -76 196 0 15 31 17 60 15 243 l-3 209 -33 29 c-26 23 -41 29 -80 29 -41 0 -53 -5 -78 -31z\"\/><path d=\"M3085 3251 c-45 -19 -58 -50 -96 -229 -47 -217 -49 -260 -13 -295 52 -53 146 -42 177 20 16 31 87 366 87 410 0 70 -86 122 -155 94z\"\/><path d=\"M1751 3234 c-13 -9 -29 -31 -37 -50 -12 -29 -10 -49 21 -204 19 -94 39 -189 45 -210 14 -50 54 -80 110 -80 34 0 48 6 76 34 21 21 34 44 34 59 0 14 -18 113 -40 219 -37 178 -43 195 -70 221 -36 32 -101 37 -139 11z\"\/><path d=\"M1163 3073 c-36 -7 -73 -59 -73 -102 0 -56 133 -378 171 -413 34 -32 83 -37 129 -13 70 36 67 87 -16 290 -86 209 -89 214 -129 231 -35 14 -42 15 -82 7z\"\/><path d=\"M3689 3066 c-15 -9 -33 -30 -42 -48 -48 -103 -147 -355 -147 -375 0 -98 131 -148 192 -74 13 15 57 108 97 206 80 196 84 226 37 273 -30 30 -99 39 -137 18z\"\/><path d=\"M583 2784 c-38 -19 -67 -74 -58 -113 9 -42 211 -354 242 -373 16 -10 45 -18 66 -18 51 0 107 52 107 100 0 39 -1 41 -124 234 -80 126 -108 162 -133 173 -41 17 -61 16 -100 -3z\"\/><path d=\"M4250 2784 c-14 -9 -74 -91 -133 -183 -95 -150 -107 -173 -107 -213 0 -55 33 -94 87 -104 67 -13 90 8 211 198 130 202 137 225 78 284 -27 27 -42 34 -72 34 -22 0 -50 -8 -64 -16z\"\/><path d=\"M2275 2693 c-553 -48 -1095 -270 -1585 -649 -135 -104 -459 -423 -483 -476 -23 -49 -22 -139 2 -186 73 -142 361 -457 571 -626 285 -228 642 -407 990 -497 242 -63 336 -73 660 -74 310 0 370 5 595 52 535 111 1045 392 1455 803 122 121 250 273 275 326 19 41 19 137 0 174 -41 79 -309 363 -465 492 -447 370 -946 591 -1479 653 -113 14 -422 18 -536 8z m395 -428 c171 -34 330 -124 456 -258 112 -119 167 -219 211 -378 27 -96 24 -300 -5 -401 -72 -255 -236 -447 -474 -557 -132 -62 -201 -76 -368 -76 -167 0 -236 14 -368 76 -213 98 -373 271 -451 485 -162 444 86 934 547 1084 153 49 292 57 452 25z m909 -232 c222 -123 408 -262 593 -441 76 -74 138 -139 138 -144 0 -16 -233 -242 -330 -319 -155 -123 -309 -223 -461 -299 l-81 -41 32 46 c18 26 49 83 70 128 143 306 141 649 -6 957 -25 52 -61 116 -79 142 l-34 47 45 -20 c26 -10 76 -36 113 -56z m-2057 25 c-40 -58 -105 -190 -130 -263 -110 -324 -59 -707 132 -981 25 -35 42 -64 37 -64 -19 0 -241 119 -326 174 -188 122 -406 314 -532 468 l-58 71 108 103 c185 178 428 349 672 473 66 33 121 60 123 61 2 0 -10 -19 -26 -42z\"\/><path d=\"M2375 1950 c-198 -44 -350 -190 -395 -379 -18 -76 -8 -221 19 -290 114 -284 457 -406 731 -260 98 52 188 154 231 260 27 69 37 214 19 290 -38 163 -166 304 -326 360 -67 23 -215 33 -279 19z\"\/><\/g><\/svg><\/i> <img loading=\"lazy\" decoding=\"async\" width=\"16\" height=\"16\" alt=\"Loading\" src=\"https:\/\/aif.amtbbs.org\/wp-content\/plugins\/page-views-count\/ajax-loader-2x.gif\" border=0 \/><\/p>\n<div class=\"pvc_clear\"><\/div>\n","protected":false},"author":56,"featured_media":23661,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[23,20,80],"tags":[1498,919,113],"class_list":["post-23659","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-23","category-20","category-80","tag-o3-mini","tag-openai","tag-113"],"_links":{"self":[{"href":"https:\/\/aif.amtbbs.org\/index.php\/wp-json\/wp\/v2\/posts\/23659","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/aif.amtbbs.org\/index.php\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/aif.amtbbs.org\/index.php\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/aif.amtbbs.org\/index.php\/wp-json\/wp\/v2\/users\/56"}],"replies":[{"embeddable":true,"href":"https:\/\/aif.amtbbs.org\/index.php\/wp-json\/wp\/v2\/comments?post=23659"}],"version-history":[{"count":1,"href":"https:\/\/aif.amtbbs.org\/index.php\/wp-json\/wp\/v2\/posts\/23659\/revisions"}],"predecessor-version":[{"id":23665,"href":"https:\/\/aif.amtbbs.org\/index.php\/wp-json\/wp\/v2\/posts\/23659\/revisions\/23665"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/aif.amtbbs.org\/index.php\/wp-json\/wp\/v2\/media\/23661"}],"wp:attachment":[{"href":"https:\/\/aif.amtbbs.org\/index.php\/wp-json\/wp\/v2\/media?parent=23659"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/aif.amtbbs.org\/index.php\/wp-json\/wp\/v2\/categories?post=23659"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/aif.amtbbs.org\/index.php\/wp-json\/wp\/v2\/tags?post=23659"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}