Analyze the performance of artificial intelligences by comparing the methodologies and results from BDM. Each tool, from ChatGPT to Le Chat, presents specific strengths and undeniable limitations. The ability to access real-time information has become an essential requirement in a constantly evolving world. The relevance of sources and the accuracy of data are also crucial. Here, the plethora of options requires in-depth analysis to navigate wisely among the various offers available in the technological market.
Detailed analysis of generative AI tools
For the past two years, artificial intelligence (AI) generative tools have been competing in ingenuity to stand out in the market. Currently, their features tend to standardize, raising the question: do these tools really offer the same level of performance? The answer emerges from a thorough comparison conducted by BDM, contrasting the performances of ChatGPT, Gemini, Claude, Perplexity, Copilot, DeepSeek, and Le Chat.
Evolution of real-time AI capabilities
Since 2023, generative AI models have shown significant advancement in real-time access to information. Tools like ChatGPT, Gemini, Claude, and Le Chat can now query the web to provide enriched and verifiable answers. This recent development aims to produce up-to-date responses and limit hallucinations or inaccurate information. BDM conducted a test by submitting three prompts focused on unusual news topics.
Testing responses on current topics
Analysis of the prompt regarding Instagram
The first prompt, related to the novelty of Instagram launched in France in October 2025, generated varied results. ChatGPT, thanks to its Web Search feature, was able to easily identify the Instagram Map. Its approach was based on reliable sources, such as Meta’s newsrooms and recognized media outlets. Its effectiveness was reflected in a clear response, accompanied by relevant data.
Performance on WhatsApp user data
The second prompt inquired about the number of active users of WhatsApp for 2025. ChatGPT provided precise details, relying on official information and articles from news agencies. However, it limited its analysis to Europe and the United States due to a lack of data for other regions, which represents a notable limitation.
Verification of false information
The last prompt required verification of false information concerning Insurance for Clio 5. Again, ChatGPT was diligent, taking over a minute to examine around fifty diverse contents. The delivered response, clear and concise, demonstrated a sustained ability to analyze and curate information.
Assessment of competitors
Performance of Gemini
Gemini showed signs of weakness during the tests. Although it has access to the Google search engine, it failed to identify the Instagram Map. Its major flaw remains a lack of transparency regarding sources, which undermines its credibility. Nevertheless, its responses are often structured and precise, and it shows considerable speed.
Effectiveness of Claude
Claude, a recent player in the market, has distinguished itself by its ability to recognize its limits. When it does not find information, it continues its search. Its speed and clarity in presenting results are also commendable. However, it sometimes relies on unique sources, which limits the veracity of the communicated information.
Perplexity: Reliability and extension of the discussion
Perplexity asserts itself as an effective hybrid model, positioned between a chatbot and a search engine. With its Quick Search feature, it easily answered queries, using diverse and relevant sources. Perplexity stands out for its ability to extend the dialogue, proposing enriching follow-up questions for the user.
Copilot and DeepSeek under examination
Copilot, based on Bing, demonstrates notable speed, but its superficiality in responses is problematic. It often limits itself to a handful of sources. DeepSeek, on the other hand, functions differently depending on the manual activation of its search feature. When activated, its responses benefit from a good level of articulation and context.
Le Chat and its speed
Le Chat has improved its search activation, displaying significant speed on simple queries. Its responses, while succinct, sometimes lack depth and rely on a limited number of sources, raising questions about the reliability of its information.
Confronting performances
The main conclusions of this analysis highlight the strengths and weaknesses of each tool. The ability to provide reliable information, to cite diverse sources, and to structure responses remains essential to gaining users’ trust. Future developments of generative AI tools must address challenges of accuracy and transparency in a constantly evolving information environment.
References
- Comparative analysis of token costs for LLMs
- Is Europe really lagging behind in the AI race?
- The limits of AI language models
- Dive into the geopolitics of generative AI
- Best AI-based image generators in 2025
Frequently asked questions on the comparative analysis of artificial intelligences
What are the main differences between ChatGPT and Gemini in web search?
ChatGPT has a web search feature that allows it to access the most recent information, while Gemini, although historically integrated with Google, has shown difficulties in sorting through recent announcements and lacks transparency about its sources.
How does Claude distinguish itself from other tools in data analysis?
Claude has a unique incremental approach: when it does not immediately find the information, it acknowledges its failure and conducts a more targeted search, allowing it to obtain a relevant response.
What criteria did BDM use to evaluate the performances of AI tools?
BDM evaluated the ability of the tools to access recent information, identify and cite reliable sources, verify information, structure responses, and provide numerical data.
Does Perplexity stand out from other tools in terms of response relevance?
Yes, Perplexity excels in the reliability of its responses. During tests, it was able to correctly identify topics and systematically cites sources, providing relevant contextual elements.
Why do some AIs like DeepSeek require manual activation of web search?
DeepSeek lowers costs by disabling web search by default, requiring it to rely on a manual method, allowing the user to choose when to activate this feature.
How does the response speed of Microsoft Copilot compare to others?
Microsoft Copilot is quick in processing queries, but its responses may lack depth and detail, often relying on a limited number of sources.
Does Le Chat provide detailed and precise responses in its analyses?
Le Chat can give quick answers, but these sometimes lack depth and rely on a limited number of sources, which may harm the reliability of the information.
What are the advantages of using ChatGPT for current news responses?
ChatGPT, thanks to its web search feature, can provide validated responses enriched with official sources, striving to deliver up-to-date and verifiable information, thus reducing hallucinations.
