ChatGPT vs Gemini Deep Research: Which AI Agent Is Better?

OpenAI and Google are currently locked in the race to offer the best Deep Research agent. Gemini’s Deep Research agent is powered by the new Gemini 2.0 Flash Thinking model whereas ChatGPT’s Deep Research agent uses a fine-tuned version of the o3 model. The interesting to note here is that Google offers its agent for free while OpenAI charges $20. Nevertheless, we’ve put both Google and OpenAI’s Deep Research agents to the test, so let’s check out how they stack up.

Task #1: Research on China’s Emergence in AI

To test the Deep Research agent in ChatGPT and Gemini, I asked them to conduct in-depth research on China’s emergence in the AI field. I asked both agents to analyze leading AI companies in China, what the government policies are, competition with US labs, and more.

ChatGPT Deep Research agent took 10 minutes to complete the task and cited 30 different sources, resulting in a 9,000-word report. On the other hand, Gemini’s Deep Research agent researched for 8 minutes and generated a much smaller 3,000-word report. However, it analyzed more than 170 websites, which is amazing.

As for the research output, I carefully reviewed both reports and found that ChatGPT’s Deep Research agent entirely failed to mention China’s recent AI advancements. It didn’t even mention DeepSeek R1, Baidu’s new Ernie 4.5 model, and more.

As it turns out, ChatGPT’s Deep Research agent relied heavily on a Stanford article and a Wikipedia page which were published and last updated in 2017, and 2021, respectively. As a result, most of the information was outdated. It didn’t even touch on the latest video generation models and robotics companies.

Gemini’s Deep Research agent, while being succinct, mentioned, “Notably, DeepSeek, a startup established in 2023, has quickly risen to prominence with its R1 model.” In addition, Gemini also wrote, “DeepSeek has also distinguished itself through its pioneering work in developing novel architectural advancements, such as Multi-Head Latent Attention (MLA).“

It also discussed video generation AI tools such as Kling AI, MiniMax, and AI-driven robotics companies like Unitree. My assessment is that Gemini’s advantage is Google’s search index, which is updated regularly with new web pages on every topic. OpenAI is likely relying on Bing to curate webpages for research work, and as a result, it’s curating old information.

Task #2: Research on The Future of AI Chips over Nvidia GPUs

In the next task, I asked the Deep Research agent on ChatGPT and Gemini to research the future of specialized AI chips over Nvidia GPUs that are currently employed for training and more. ChatGPT included 22 sources and generated a large document with tables and in-line citations. Gemini, just like earlier, analyzed over 100 web pages but created a short document.

Both agents mentioned the rise of TPUs, ASICs, FPGAs, LPUs, and WSEs, and reported Nvidia’s CUDA moat. They discussed new silicon companies like Groq, Cerebras, SambaNova, Graphcore, and others as well. ChatGPT’s Deep Research AI agent also touched on Huawei’s Ascend AI chip, which Gemini missed.

ChatGPT Deep Research also mentioned Amazon’s Trainium and Inferentia chips, Microsoft’s MAIA chip, Intel’s Gaudi accelerator, and Google’s TPU. However, much like its previous fumble, ChatGPT is stuck in the past. For example, it has no information that Google has already released the TPU v6e (Trillium) chip. Next, ChatGPT thinks Nvidia’s Blackwell B200 is an upcoming GPU, but it was released in 2024.

Despite its misses, I would give it to ChatGPT’s Deep Research agent as the report has more depth and insights than Gemini. That said, it’s important to note that there is a fundamental issue with ChatGPT’s Deep Research agent.

The agent is powered by a fine-tuned version of OpenAI’s o3 model, which has a training cut-off date of October 2023. As a result, the internal model doesn’t have the latest knowledge. Despite using the internet for current information, it often misses out on the latest advancements.

Task #3: Research on Obesity Management in 2025

In the third task, I asked the Deep Research agent on ChatGPT and Gemini to research obesity management in 2025. I specifically asked both agents to include the latest advancements in 2025, covering all kinds of therapies. ChatGPT’s Deep Research cited high-quality sources and medical journals to explain the current therapy in obesity management.

It mentioned new breakthroughs like GLP-1 peptides, their side effects, FDA approval status of new drugs, and even gene therapy. Overall, ChatGPT underlined the current obesity management policy of the US, UK, and Canada.

On the other hand, Gemini’s Deep Research agent also did a wonderful job. It listed all the medical interventions and included upcoming drug trials in a tabular format. What I found interesting is that Gemini mentioned novel drugs like the triple hormone receptor (Retatrutide), which is creating a buzz in the medical field for significant weight loss.

In addition, Googlr’d Gemini dived deep into companies like Novo Nordisk and Eli Lilly, which are experimenting with novel drugs for weight reduction, giving a complete picture of the development. All in all, I would say that Gemini’s generated report was well-structured and included the latest information on obesity management.

Should You Use ChatGPT Deep Research or Gemini Deep Research?

In my testing, I preferred Gemini’s Deep Research AI agent because it consistently included current information on various topics. While ChatGPT’s Deep Research agent covers a broad range of ideas, its limited knowledge of the latest development, hinders it from giving a complete picture. On top of that, Gemini’s Deep Research AI agent is free for all users while OpenAI charges $20 to access its agent.

The reason Gemini has gotten so better is that it’s now powered by the Gemini 2.0 Flash Thinking model. Compared to the previous Gemini 1.5 Pro model, the new reasoning model thinks and plans what information it needs to complete the research work.

That said, one big advantage of ChatGPT’s Deep Research agent is that you can upload files whereas Gemini doesn’t allow file uploads while using its agent. This can be particularly helpful in STEM-related research work where you can upload PDFs of paywalled science journals.

Since many of these journals are behind a paywall and inaccessible via web search, you can upload files manually to provide a richer knowledge base. This will allow the agent to incorporate valuable insights. However, Gemini lets you export the generated report to Google Docs, which is helpful for sharing.

All in all, I suggest using Gemini’s Deep Research AI agent before subscribing to the ChatGPT Plus plan. Google has improved the agent significantly, and it’s much faster now.