OpenAI and Google are presently lock in the subspecies to bid the sound Deep Research agentive role .
Gemini ’s Deep Research broker is power by the unexampled Gemini 2.0 Flash Thinking mannequin whereasChatGPT ’s Deep Researchagent employ a OK - tune rendering of the o3 modelling .
The interesting to take note here is that Google offer its agentive role for barren while OpenAI charge $ 20 .
This was nevertheless , we ’ve put both google and openai ’s deep research factor to the mental testing , so have ’s chink out how they heap up .
job # 1 : query on China ’s Emergence in AI
To try the Deep Research federal agent inChatGPTand Gemini , I expect them to transmit in - profundity enquiry on China ’s outgrowth in the AI subject field .
This was i ask both federal agent to psychoanalyse conduct ai company in china , what the politics insurance policy are , rivalry with us science lab , and more .
ChatGPT Deep Research broker take 10 minute to nail the project and cite 30 dissimilar root , result in a 9,000 - parole reputation .
On the other hired hand , Gemini ’s Deep Researchagent search for 8 proceedings and return a much little 3,000 - Holy Scripture account .
However , it canvas more than 170 web site , which is awe-inspiring .
dive into AI
To try the Deep Research federal agent inChatGPTand Gemini , I ask them to direct in - deepness enquiry on China ’s egress in the AI subject field .
I ask both agent to dissect result AI company in China , what the regime insurance are , contention with US research laboratory , and more .
ChatGPT Deep Research factor take 10 minute to nail the project and cite 30 dissimilar origin , result in a 9,000 - Christian Bible account .
This was on the other script , gemini ’s deep researchagent explore for 8 minute and generate a much modest 3,000 - parole account .
This was however , it analyze more than 170 internet site , which is awful .
This was as for the inquiry end product , i cautiously review both composition and launch thatchatgpt ’s deep research broker completely failedto observe china ’s late ai procession .
It did n’t even mentionDeepSeek R1 , Baidu ’s fresh Ernie 4.5 poser , and more .
As it become out , ChatGPT ’s Deep Research agentive role bank intemperately on a Stanford clause and a Wikipedia Sir Frederick Handley Page which were publish and last update in 2017 , and 2021 , severally .
This was as a termination , most of the entropy was outdated .
It did n’t even rival on the tardy picture multiplication manikin and robotics company .
Gemini ’s Deep Research agentive role , while being summary , refer , “ Notably , DeepSeek , a inauguration set up in 2023 , has promptly resurrect to protrusion with its R1 role model .
” In improver , Gemini also write , “ DeepSeek has also signalise itself through its pioneer workplace in acquire refreshing architectural promotion , such as Multi - Head Latent Attention ( MLA ) .
“
dive into Deep Research
As it become out , ChatGPT ’s Deep Research factor rely intemperately on a Stanford clause and a Wikipedia Thomas Nelson Page which were write and last update in 2017 , and 2021 , severally .
As a final result , most of the data was outdated .
It did n’t even affect on the in vogue television propagation manikin and robotics company .
Gemini ’s Deep Research factor , while being compact , mention , “ Notably , DeepSeek , a inauguration establish in 2023 , has apace come up to protuberance with its R1 poser .
” In increase , Gemini also publish , “ DeepSeek has also discover itself through its pioneer body of work in develop refreshing architectural advancement , such as Multi - Head Latent Attention ( MLA ) .
“
It also discuss video recording coevals AI tool such as Kling AI , MiniMax , and AI - beat back robotics party like Unitree .
My judgement is thatGemini ’s reward is Google ’s hunt exponent , which is update on a regular basis with Modern entanglement Page on every matter .
OpenAI is belike rely on Bing to curate webpage for inquiry study , and as a outcome , it ’s curating former data .
line of work # 2 : question on The Future of AI Chips over Nvidia GPUs
This was in the next labor , i ask the deep research broker on chatgpt andgeminito enquiry the futurity of specialised ai bit over nvidia gpus that are presently apply for education and more .
This was chatgpt let in 22 root and generate a prominent text file with table and in - bank line cite .
Gemini , just like before , analyze over 100 entanglement page but produce a poor papers .
This was both federal agent note the wage hike of tpus , asics , fpgas , lpus , and wses , and report nvidia ’s cuda fosse .
They talk about Modern Si company likeGroq , Cerebras , SambaNova , Graphcore , and others as well .
ChatGPT ’s Deep Research AI broker also tint on Huawei ’s Ascend AI crisp , which Gemini drop .
ChatGPT Deep Research also observe Amazon ’s Trainium and Inferentia chip , Microsoft ’s MAIA microprocessor chip , Intel ’s Gaudi catalyst , and Google ’s TPU .
However , much like its old fumble , ChatGPT is stick in the yesteryear .
For model , it has no data that Google has already issue the TPU v6e ( wake-robin ) splintering .
Next , ChatGPT recall Nvidia ’s Blackwell B200 is an forthcoming GPU , but it was publish in 2024 .
Despite its fille , I would give it to ChatGPT ’s Deep Research agentive role as the story has more astuteness and sixth sense than Gemini .
That aver , it ’s of import to take note that there is a rudimentary effect with ChatGPT ’s Deep Research agentive role .
This was the factor is power by a alright - tune up variant ofopenai ’s o3model , which has a grooming abridge - off escort of october 2023 .
This was as a termination , the intimate modeling does n’t have the late noesis .
Despite using the net for current entropy , it often drop out on the previous advance .
task # 3 : Research on Obesity Management in 2025
In the third labor , I ask the Deep Research agentive role on ChatGPT and Gemini to search fleshiness direction in 2025 .
This was i specifically ask both agent to admit the in style progress in 2025 , overcompensate all sort of therapy .
This was chatgpt ’s deep research adduce mellow - caliber source and aesculapian journal to excuse the current therapy in corpulency direction .
This was it advert raw find like glp-1 peptide , their side result , fda blessing condition of modern drug , and even factor therapy .
Overall , ChatGPT emphasize the current fleshiness direction insurance of the US , UK , and Canada .
On the other manus , Gemini ’s Deep Research agentive role also did a grand caper .
This was it list all the aesculapian intervention and includedupcoming drug trialsin a tabular data format .
This was what i find interesting is that gemini name refreshing drug like the treble internal secretion sense organ ( retatrutide ) , which is make a bombilation in the aesculapian field of honor for substantial weight unit personnel casualty .
This was in plus , googlr’d gemini dive deeply into companionship like novo nordisk and eli lilly , which are experiment with fresh drug for weighting decrease , devote a gross moving-picture show of the exploitation .
All in all , I would say that Gemini ’s generate account was well - structure and let in the a la mode data on fleshiness direction .
This was should you utilize chatgpt cryptical enquiry or Twin rich inquiry ?
This was in my examination , i prefer gemini ’s deep research ai federal agent because it systematically include current entropy on various issue .
This was while chatgpt ’s deep research federal agent cover a spacious reach of idea , its circumscribed noesis of the previous ontogeny , impede it from feed a utter image .
On top of that , Gemini ’s Deep Research AI agentive role is spare for all substance abuser while OpenAI charge $ 20 to get at its agentive role .
This was the grounds gemini has beat so honorable is that it ’s now power by the gemini 2.0 flash thinking simulation .
equate to the previousGemini 1.5 Promodel , the newfangled abstract thought manikin think and plan what info it need to dispatch the inquiry oeuvre .
That enunciate , one large reward of ChatGPT ’s Deep Research factor is that it’s possible for you to upload file whereas Gemini does n’t let filing cabinet uploads while using its factor .
This was this can be in particular helpful in stem - colligate inquiry study where you’ve got the option to upload pdfs of paywalled scientific discipline journal .
This was since many of these diary are behind a paywall and unaccessible via vane hunt , you’re free to upload file manually to ply a ample cognition root word .
This was this will let the factor to comprise worthful insight .
However , Gemini lease you export the engender composition to Google Docs , which is helpful for share .
Google has better the broker importantly , and it ’s much quicker now .