گہری تحقیق کا تعارف
ایک ایجنٹ جو ریزننگ کا استعمال کرتے ہوئے بڑی مقدار میں آن لائن معلومات کو یکجا کرے اور آپ کے لیے کئی مراحل پر مشتمل ریسرچ ٹاسکس مکمل کرے۔ آج سے Pro صارفین کے لیے دستیاب ہے، جبکہ Plus اور Team کے لیے بعد میں دستیاب ہوگا۔
10 فروری، 2026 پر اپ ڈیٹ: اب آپ ڈیپ ریسرچ کو کسی بھی MCP یا ایپ سے منسلک کر سکتے ہیں اور ویب تلاش کو قابل اعتماد سائٹس تک محدود کر سکتے ہیں، تاکہ آپ مستند، صنعت کے معیاری ذرائع پر توجہ مرکوز کر سکیں. آپ اب حقیقی وقت میں پیش رفت کو بھی ٹریک کر سکتے ہیں اور فالو اپ پرومپٹس یا نئے ذرائع کے ساتھ بہتر بنانے کے لیے مداخلت کر سکتے ہیں. ہم نے بصری تجربہ کو اپ ڈیٹ کر دیا ہے تاکہ شروع کرنا، ٹریک کرنا اور اپنی تحقیق کو ابتدا سے انتہا تک جائزہ لینا زیادہ آسان ہو.
17 جولائی 2025 اپڈیٹ: ڈیپ ریسرچ اب ChatGPT ایجنٹ کا حصہ بننے والے ایک بصری براؤزر تک رسائی کے ذریعے مزید گہرائی اور وسعت اختیار کر چکا ہے۔ ان اپڈیٹڈ صلاحیتوں تک رسائی کے لیے بس کمپوزر میں ڈراپ ڈاؤن سے “agent mode” منتخب کریں اور اپنی کیوری درج کریں۔ ڈیپ ریسرچ کی اصل فنکشنلٹی اب بھی ٹولز مینو میں “deep research” آپشن کے ذریعے دستیاب ہے۔
24 اپریل 2025 اپڈیٹ: ہم اس بات میں نمایاں اضافہ کر رہے ہیں کہ آپ ڈیپ ریسرچ کو کتنی بار استعمال کر سکتے ہیں—Plus، Team، Enterprise، اور Edu صارفین کو اب ماہانہ 25 کیوریز ملتی ہیں، Pro صارفین کو 250، اور Free صارفین کو 5۔ یہ ایک نئے لائٹ ویٹ ورژن کے ذریعے ممکن ہوا ہے جو o4-mini کے ایک ورژن سے چلتا ہے، جسے زیادہ cost-efficient بناتے ہوئے اعلیٰ معیار برقرار رکھا گیا ہے۔ جب آپ فل ورژن کی حد تک پہنچ جائیں گے تو آپ کی کیوریز خودکار طور پر لائٹ ویٹ ورژن پر منتقل ہو جائیں گی۔
25 فروری 2025 اپڈیٹ: تمام Plus صارفین اب ڈیپ ریسرچ استعمال کر سکتے ہیں۔
5 فروری 2025 اپڈیٹ: ڈیپ ریسرچ اب برطانیہ، سوئٹزرلینڈ، اور European Economic Area کے Pro صارفین کے لیے دستیاب ہے۔
آج ہم ChatGPT میں ڈیپ ریسرچ متعارف کر رہے ہیں—ایک نئی ایجنٹک صلاحیت جو پیچیدہ ٹاسکس کے لیے انٹرنیٹ پر ملٹی-اسٹیپ ریسرچ کرتی ہے۔ یہ چند منٹوں میں وہ کچھ مکمل کر لیتا ہے جن پر کسی انسان کو کئی گھنٹے لگیں۔
ڈیپ ریسرچ OpenAI کا اگلا ایجنٹ ہے جو آپ کے لیے خودمختار طور پر کام انجام دے سکتا ہے—آپ اسے کوئی پرامپٹ دیں، اور ChatGPT سینکڑوں آن لائن سورسز کو تلاش، تجزیہ، اور یکجا کرکے ایک جامع رپورٹ تیار کرتا ہے جو ایک ریسرچ اینالسٹ کے معیار کی ہوتی ہے۔ یہ آنے والے OpenAI o3 ماڈل کے ایک ایسے ورژن سے چلتا ہے جو ویب براؤزنگ اور ڈیٹا اینالسس کے لیے آپٹیمائزڈ ہے۔ یہ ریزننگ استعمال کرتے ہوئے انٹرنیٹ پر موجود بےشمار ٹیکسٹ، تصاویر، اور PDFs کو تلاش، سمجھ اور تجزیہ کرتا ہے—اور سامنے آنے والی معلومات کے مطابق اپنا رخ بدلتا رہتا ہے۔
علم کو یکجا کرنے کی صلاحیت نئے علم تخلیق کرنے کی بنیادی شرط ہے۔ اسی وجہ سے، ڈیپ ریسرچ ہمارا یہ وسیع تر ہدف حاصل کرنے کی سمت ایک اہم قدم ہے کہ ہم ایسا AGI تیار کریں جو نئے سائنسی ریسرچ پیدا کرنے کے قابل ہو۔
Deep research is built for people who do intensive knowledge work in areas like finance, science, policy, and engineering and need thorough, precise, and reliable research. It can be equally useful for discerning shoppers looking for hyper-personalized recommendations on purchases that typically require careful research, like cars, appliances, and furniture. Every output is fully documented, with clear citations and a summary of its thinking, making it easy to reference and verify the information. It is particularly effective at finding niche, non-intuitive information that would require browsing numerous websites. Deep research frees up valuable time by allowing you to offload and expedite complex, time-intensive web research with just one query.
Deep research independently discovers, reasons about, and consolidates insights from across the web. To accomplish this, it was trained on real-world tasks requiring browser and Python tool use, using the same reinforcement learning methods behind OpenAI o1, our first reasoning model. While o1 demonstrates impressive capabilities in coding, math, and other technical domains, many real-world challenges demand extensive context and information gathering from diverse online sources. Deep research builds on these reasoning capabilities to bridge that gap, allowing it to take on the types of problems people face in work and everyday life.
In ChatGPT, select ‘deep research’ in the message composer and enter your query. Tell ChatGPT what you need—whether it’s a competitive analysis on streaming platforms or a personalized report on the best commuter bike. You can attach files or spreadsheets to add context to your question. Once it starts running, a sidebar appears with a summary of the steps taken and sources used.
Deep research may take anywhere from 5 to 30 minutes to complete its work, taking the time needed to dive deep into the web. In the meantime, you can step away or work on other tasks—you’ll get a notification once the research is complete. The final output arrives as a report within the chat – in the next few weeks, we will also be adding embedded images, data visualizations, and other analytic outputs in these reports for additional clarity and context.
Compared to deep research, GPT‑4o is ideal for real-time, multimodal conversations. For multi-faceted, domain-specific inquiries where depth and detail are critical, deep research’s ability to conduct extensive exploration and cite each claim is the difference between a quick summary and a well-documented, verified answer that can be usable as a work product.
GPT-4o
Deep research
Deep research responds to the prompt in a highly detailed manner, providing side-by-side country-specific data for both top 10 developed and top 10 developing countries for easy reference and comparison. It uses that information to offer detailed market-entry recommendations that are informed and usable.
ڈیپ ریسرچ کو متعدد ڈومینز میں مشکل براؤزنگ اور ریزننگ ٹاسکس پر end-to-end reinforcement learning کے ذریعے ٹرین کیا گیا۔ اس ٹریننگ کے ذریعے، اس نے مطلوبہ ڈیٹا تلاش کرنے کے لیے ایک ملٹی-اسٹیپ راستہ پلان اور ایکزیکیوٹ کرنا سیکھا، جہاں ضرورت پڑنے پر وہ واپس پلٹ کر حقیقی وقت کی معلومات کے مطابق ردعمل دے سکتا ہے۔ یہ ماڈل صارف کی اَپ لوڈ کی گئی فائلز پر بھی براؤز کر سکتا ہے، python ٹول استعمال کرتے ہوئے گراف بنا اور تبدیل کر سکتا ہے، پیدا کیے گئے گراف اور ویب سائٹس سے حاصل شدہ تصاویر کو اپنے رسپانسز میں ایمبیڈ کر سکتا ہے، اور اپنے سورسز کے مخصوص جملے یا پیراگراف کو cite کر سکتا ہے۔ اس ٹریننگ کے نتیجے میں، یہ حقیقی دنیا کے مسائل پر مرکوز کئی عوامی ایویلوئیشنز میں نئی بلندیاں حاصل کرتا ہے۔
Humanity’s Last Exam(نئی ونڈو میں کھلتا ہے) پر، جو حال ہی میں جاری کی گئی ایک ایویلیوئیشن ہے جس میں مختلف مضامین کے ماہر سطح کے سوالات پر AI کی جانچ ہوتی ہے، ڈیپ ریسرچ کو پاور کرنے والا ماڈل 26.6% درستگی کے ساتھ ایک نیا بلند ترین اسکور حاصل کرتا ہے۔ اس ٹیسٹ میں 3,000 سے زائد ملٹی پل چوائس اور مختصر جوابی سوالات شامل ہیں، جو 100 سے زیادہ مضامین پر مشتمل ہیں—لسانیات سے لے کر راکٹ سائنس تک، کلاسکس سے ماحولیاتیات تک۔ OpenAI o1 کے مقابلے میں سب سے زیادہ بہتری کیمسٹری، ہیومینیٹیز اور سوشل سائنسز اور ریاضی میں نظر آئی۔ ڈیپ ریسرچ کو پاور کرنے والے ماڈل نے ضرورت پڑنے پر مؤثر طریقے سے خصوصی معلومات تلاش کرکے انسانی طرزِ عمل جیسا رجحان ظاہر کیا۔
| ماڈل | درستگی (%) |
|---|---|
| GPT-4o | ۳٫۳ |
| Grok-2 | ۳٫۸ |
| Claude 3.5 Sonnet | ۴٫۳ |
| Gemini Thinking | ۶٫۲ |
| OpenAI o1 | ۹٫۱ |
| DeepSeek-R1* | ۹٫۴ |
| OpenAI o3-mini (medium)* | ۱۰٫۵ |
| OpenAI o3-mini (ہائی)* | ۱۳٫۰ |
| OpenAI گہری تحقیق** | ۲۶٫۶ |
GAIA(نئی ونڈو میں کھلتا ہے)1 پر—جو ایک عوامی بینچ مارک ہے جو حقیقی دنیا کے سوالات پر AI کا جائزہ لیتا ہے—ڈیپ ریسرچ کو چلانے والا ماڈل ایک نئی SOTA سطح پر پہنچتا ہے، اور بیرونی لیڈر بورڈ(نئی ونڈو میں کھلتا ہے) میں سرفہرست آتا ہے۔ تین سطحوں کی مشکل پر مشتمل ان سوالات کی کامیاب تکمیل کے لیے ریزننگ، ملٹی-موڈل روانی، ویب براؤزنگ، اور ٹول-استعمال کی مہارت جیسی صلاحیتیں درکار ہیں۔
| GAIA | ||||
|---|---|---|---|---|
| لیول 1 | لیول 2 | لیول 3 | اوسط | |
| پچھلا SOTA(نئی ونڈو میں کھلتا ہے) | ۶۷٫۹۲ | ۶۷٫۴۴ | ۴۲٫۳۱ | ۶۳٫۶۴ |
| گہری تحقیق (pass@1) | ۷۴٫۲۹ | ۶۹٫۰۶ | ۴۷٫۶ | ۶۷٫۳۶ |
| گہری تحقیق (cons@64) | ۷۸٫۶۶ | ۷۳٫۲۱ | ۵۸٫۰۳ | ۷۲٫۵۷ |
GAIA ٹاسک کی مثالیں
متعدد شعبوں میں ایکسپرٹ-لیول ٹاسکس کی ایک اندرونی ایویلوئیشن میں، ڈومین ماہرین نے ڈیپ ریسرچ کو کئی گھنٹوں کی مشکل، دستی تحقیق کو خودکار بنانے کے قابل قرار دیا۔
پاس ریٹ بمقابلہ زیادہ سے زیادہ ٹول کالز
ایکسپرٹ-لیول ٹاسک مثالیں
ماہر سطح کے ٹاسکوں میں کامیابی کی شرح تخمینی معاشی قدر کے لحاظ سے
ماہر سطح کے ٹاسکوں میں اندازہ شدہ گھنٹوں کے مطابق کامیابی کی شرح
Deep research unlocks significant new capabilities, but it’s still early and has limitations. It can sometimes hallucinate facts in responses or make incorrect inferences, though at a notably lower rate than existing ChatGPT models, according to internal evaluations. It may struggle with distinguishing authoritative information from rumors, and currently shows weakness in confidence calibration, often failing to convey uncertainty accurately. At launch, there may be minor formatting errors in reports and citations, and tasks may take longer to kick off. We expect all these issues to quickly improve with more usage and time.
Deep research in ChatGPT is currently very compute intensive. The longer it takes to research a query, the more inference compute is required. We are starting with a version optimized for Pro users today, with up to 100 queries per month. Plus and Team users will get access next, followed by Enterprise. We are still working on bringing access to users in the United Kingdom, Switzerland, and the European Economic Area.
All paid users will soon get significantly higher rate limits when we release a faster, more cost-effective version of deep research powered by a smaller model that still provides high quality results.
In the coming weeks and months, we’ll be working on the technical infrastructure, closely monitoring the current release, and conducting even more rigorous testing. This aligns with our principle of iterative deployment. If all safety checks continue to meet our release standards, we anticipate releasing deep research to Plus users in about a month.
Deep research آج سے ChatGPT ویب پر دستیاب ہے، اور اس ماہ کے اندر موبائل اور ڈیسک ٹاپ ایپس پر بھی جاری کر دیا جائے گا۔ اس وقت deep research اوپن ویب اور اپ لوڈ کی گئی کسی بھی فائل تک رسائی حاصل کر سکتا ہے۔ مستقبل میں، آپ مزید خصوصی ڈیٹا سورسز سے کنیکٹ کر سکیں گے—جس سے اس کی رسائی سبسکرپشن پر مبنی یا اندرونی ریسورسز تک وسیع ہو جائے گی—تاکہ اس کا آؤٹ پٹ مزید مضبوط اور ذاتی نوعیت کا ہو سکے۔
مزید آگے دیکھتے ہوئے، ہم ایسے ایجنٹک تجربات کا تصور کرتے ہیں جو حقیقی دنیا کی غیرہم وقتی تحقیق اور عمل درآمد کے لیے ChatGPT میں یکجا ہوں گے۔ ڈیپ ریسرچ، جو غیرہم وقتی آن لائن تحقیق انجام دے سکتا ہے، اور Operator، جو حقیقی دنیا میں ایکشن لے سکتا ہے، کے امتزاج سے ChatGPT آپ کے لیے بتدریج زیادہ جدید نوعیت کے ٹاسکس انجام دے سکے گا۔
3 فروری 2025 کا اضافہ: ہم نے o3 کے ابتدائی ورژن — جو deep research کو پاور کرتا ہے — پر سخت حفاظتی ٹیسٹنگ، تیاری کی جانچ اور گورننس ریویوز مکمل کیے، اور اسے Medium(نئی ونڈو میں کھلتا ہے) رسک کے طور پر شناخت کیا۔ ہم نے اضافی حفاظتی ٹیسٹنگ بھی انجام دی تاکہ deep research کی ویب براؤزنگ صلاحیت سے وابستہ اضافی خطروں کو بہتر طور پر سمجھا جا سکے، اور ہم نے نئی حفاظتی اقدامات شامل کیے ہیں۔ ہم موجودہ محدود ریلیز کی بھرپور جانچ اور مسلسل نگرانی جاری رکھیں گے۔ جب ہم Plus صارفین کے لیے اس تک رسائی وسیع کریں گے تو deep research سے متعلق اپنے حفاظتی نتائج اور حفاظتی اقدامات ایک سسٹم کارڈ کے ذریعے جاری کریں گے۔
Footnotes
- 1
We found that the ground-truth answers for this dataset were widely leaked online and have blocked several websites or URLs accordingly to ensure a fair evaluation of the model.
Authors
Research Leads
Isa Fulford، Zhiqing Sun
Foundational Contributors
Alex Tachard Passos، Alexandra Barr، Allison Tam، Charlotte Cole، Hyung Won Chung، Jason Wei، Jon Blackman، Scott Mayer McKinney، Valerie Qi
Core Contributors
Research
Elaine Ya Le, Eric Mitchell, Eric Wallace, Hyung Won Chung, Ignasi Clavera, Leo Liu, Lorenz Kuhn, Louis Feuvrier, Max Schwarzer, Saachi Jain, Scottie Yan, Shunyu Yao, Vitchyr Pong
Deployment
Carpus Chang, Harry Zhao, Joseph Trasatti, Joshua Dickens, Matt Kaufer, Mike Trpcic, Minnia Feng, Neel Ajjarapu, Peter Vidani, Sean Fitzgerald
Contributors
Research
Ahmed El-Kishky, AJ Ostrow, Alexander Wei, Andrei Gheorghe, Andrew Kondrich, Andrey Mishchenko, Anuj Nair, Behrooz Ghorbani, Brydon Eastman, Chak Li, Foivos Tsimpourlas, Francis Song, Giambattista Parascandolo,Gildas Chabot, Hessam Bagherinezhad, Haitang Hu, Hongyu Ren, Henry Aspegren, Hunter Lightman, Ilya Kostrikov, Ilge Akkaya, James Lennon, Jean Harb, Jonathan Ward, Kai Chen, Katy Shi, Kevin Liu, Kevin Yu, Manuka Stratta, Marvin Zhang, Mengyuan Yan, Mostafa Rohaninejad, Noam Brown, Phoebe Thacker, Raz Goan, Reah Miyara, Spencer Papay, Taylor Gordon, Wenda Zhou, Wenlei Xie, Yash Patil, Yann Dubois, Youlong Cheng, Yushi Wang, Wyatt Thompson
+ all the contributors to o3.
Safety Systems
Adam Kalai, Alex Beutel, Andrea Vallone, Andy Applebaum, David Robinson, Elizabeth Proehl, Evan Mays, Grace Zhao, Irina Kofman, Jason Phang, Joaquin Quinonero Candela, Joel Parish, Kevin Liu, Kristen Ying, Lama Ahmad, Leon Maksin, Leyton Ho, Meghan Shah, Michele Wang, Miles Wang, Phillip Guo, Olivia Watkins, Owen Campbell-Moore, Patrick Chao, Sam Toizer, Samuel Miserendino, Sandhini Agarwal, Tejal Patwardhan, Tina Sriskandarajah, Troy Peterson, Yaodong Yu, Yunyun Wang
Deployment
Adam Koppel, Adam Wells, Adele Li, Andy Applebaum, Andrey Malevich, Andrew Duberstein, Andrew Howell, Anton Tananaev, Ashley Tyra, Brandon Walkin, Bryan Ashley, Cary Bassin, Cary Hudson, Cory Decareaux, Cristina Scheau, Derek Chen, Dibya Bhattacharjee, Drea Lopez, Eric Antonow, Eric Burke, Filippo Raso, Fotis Chantzis, Freddie Sulit, Harris Cohen, Heather Whitney, Jay Dixit, Jeffrey Han, Jen Robinson, Jessica Shieh, Joel Parish, Kan Wu, Kevin Gladstone, Kshitij Wadhwa, Leo Vandriel, Leyton Ho, Liang Chen, Madeline Christian, Mamie Rheingold, Matt Jones, Michelle Fradin, Mike McClay, Mingxuan Wang, Nacho Soto, Niko Felix, Patrick Delaney, Paul McMillan, Philip Pronin, Rodrigo Riaza Perez, Samuel Miserendino, Scott Ethersmith, Steven Baldwin, Thomas Dimson, Tomo Hiratsuka, Yaming Lin, Yara Khakbaz, Yining Chen
Leadership
Akshay Nathan، Greg Brockman، Hannah Wong، Jakub Pachocki، Jerry Tworek، Johannes Heidecke، Josh Tobin، Liam Fedus، Mark Chen، Mia Glaese، Nick Turley، Sam Altman، Wojciech Zaremba