More

    Gemini Pro Review: My Go-To AI Chatbot Costs $20 Per Month

    9.0/ 10 SCORE Google Gemini Pro Review Pros

    Higher-quality responses

    Accurate, with vastly lowered hallucinations

    Connection to web and to different Google companies

    Fast picture technology

    2TB cloud storage
    Cons

    Can nonetheless make logical errors

    Coding help could make repeated errors

    Failed sure obscure requests

    If the Gemini twins in Greek mythology are supposed to information sailors, then the title Google gave to its AI chatbot lastly matches its demigodly ambitions. Compared to once I examined it final yr, Gemini has seen great enhancements in accuracy and usefulness.

    While the free model of Gemini is extremely succesful and is nice for many use circumstances, the paid model I’m reviewing right here brings a extra highly effective AI mannequin that may deal with requests of upper complexity with higher “reasoning” talents. I’ve discovered responses to be extra informative, with higher nuance. Where 2.5 Flash goals to be a light-weight mannequin that may output solutions rapidly at little price, 2.5 Pro takes the additional processing time to offer higher outputs. 

    At $20 per thirty days, Gemini Pro is well worth the improve for individuals seeking to accomplish more-complex analysis and coding duties, or who desire a deeper AI mannequin to speak with. Considering {that a} Gemini Pro subscription additionally comes with 2TB of cloud storage, together with some video technology instruments, it might simply develop into a justifiable expense for some individuals.

    At the identical time, Gemini is not all-knowing. It nonetheless could make some logical errors, as a result of AI chatbots do not actually perceive the world round them. They’re extremely complicated autocomplete instruments and, consequently, can get info mistaken as a result of they do not expertise the world like we do. 

    Compared with ChatGPT Plus, Gemini Pro can nonetheless lag behind in some eventualities, regardless of its quite a few enhancements and additional integrations with different Google companies, akin to Search and Maps.

    What Google finally delivers is a extremely succesful AI chatbot that may deal with a large host of challenges. From coding to analysis, Gemini can deal with just about something thrown at it. Both ChatGPT Plus, particularly the “reasoning” o3 mannequin, and Gemini Pro supply great performance, and differentiating between the 2 comes right down to very particular use circumstances. 

    How CNET critiques AI fashions

    Last yr, I handled my critiques of AI chatbots as if I have been reviewing another tech product at CNET, operating a collection of comparability exams to see which got here out on high. Though that is a helpful solution to check digital camera high quality between the iPhone 16 Pro and the Samsung Galaxy S25 Ultra, it is rather less helpful when reviewing AI chatbots. 

    Because AI chatbots are machines that may do virtually something, performing side-by-side A-B testing is not indicative of how most individuals work together with AI. Imagine if I have been reviewing Google Search and Bing. In this situation, it might make sense to do comparative searches and file my outcomes. But AI chatbots do not work like conventional on-line search, which indexes websites to tug up essentially the most related outcomes. AIs give novel solutions to each query requested, even when it is the identical query. It makes commonplace comparative testing much less of a mirrored image of real-world use. 

    This yr, I’m choosing a extra experiential strategy. It’s much less scientific, certain, however I really feel it offers a greater sense of the vibe every AI chatbot brings. And contemplating that Google, OpenAI, Anthropic and others need to give their AI chatbots a little bit of a persona, unpacking that vibe is core to evaluating the AI expertise.

    Research and accuracy

    Compared with 2024, Gemini Pro is leagues extra correct this yr. It seldom makes up info or hyperlinks to nonexistent YouTube movies, because the earlier model did throughout my exams final yr. Google has additionally accomplished a a lot better job of integrating info gathered through Search to tug up essentially the most related sourcing. 

    To check how Gemini might assist me analysis present occasions, I requested the chatbot to investigate speaking factors from the latest New York mayoral main. It did a superb job of pulling collectively info with correct sourcing, together with Radio Free Europe, PBS, official authorities sources and, in some situations, YouTube movies from main information channels. Because Google owns YouTube, Gemini can hyperlink on to any of the location’s huge trove of movies. That offers Google an edge over the businesses behind different AI engines. YouTube, by default, blocks creators from permitting their movies for use for AI coaching. Other AI corporations have crawled YouTube to coach their fashions, a follow that violated YouTube’s phrases of service. 

    Something I’ve discovered notably helpful is utilizing Gemini as a sounding board for a few of my crazier concepts. I personal a wide array of online game consoles, all plugged into my tv with a number of HDMI switches and energy strips. I requested Gemini if it might be doable to make a superlong HDMI change that would match greater than 20 units. Gemini defined that making a circuit board that would deal with 4K, HDR and excessive refresh charges throughout a number of inputs can be extraordinarily difficult and past the scope of a DIY challenge. 

    When I requested Gemini to create a schematic of what this challenge would possibly appear like, it tried to take action with ASCII characters however finally failed. 
    Google Gemini Pro fails at making a schematic for a 24-port HDMI change. Imad Khan/CNET
    At least Gemini is actual with me.

    Despite Gemini’s accuracy, it would not perceive the world round it. I’ve just lately been making an attempt to chop extra sugar out of my weight loss plan, (apologies to my native bubble tea store). I’ve been eager to make a primary milk tea utilizing monkfruit sweetener and nondairy creamer. I requested Gemini to create me a wholesome milk tea recipe, but it surely did not work out so nicely.

    Gemini instructed I make a tea base with one cup of water and two baggage of black tea. For sweetness, Gemini mentioned so as to add solely 1 tablespoon of monkfruit sweetener. That, plus three-quarters of a cup of milk and 1 to 2 tablespoons of creamer would create that superb low-calorie milk tea. 

    The end result was a chunky mess. After I used these precise measurements, I discovered that the drink wasn’t wherever candy sufficient and that the creamer-to-liquid ratio was all off, resulting in clumps within the remaining product. Gemini has by no means sipped bubble tea earlier than, so it is sensible that it would not perceive what 2 tablespoons of nondairy creamer would do in solely 14 ounces of liquid. 

    Gemini’s ‘vibe coding’ blew me away

    I’m not a coder. Back in school, I took an introductory Python course and after a lot wrestle managed to get a C. But with AI chatbots, you do not have to be a coding wiz, and even know show “Hello World” in any respect. With Gemini Pro, I used to be capable of make a totally functioning Chrome extension with nearly no expertise (I did not check coding with Gemini Free). 
    Vibe coding is a time period that basically means to code by speaking it out. You categorical to an AI what you are hoping to perform, and thru your back-and-forth dialog, the AI will generate the code and assist you to get a program up and operating. 

    I vibe coded with Gemini, and never solely was the expertise fascinating and quite a lot of enjoyable, I additionally discovered it to be simply as impactful as once I used ChatGPT for the primary time in late 2023. 

    (Gemini is not the one AI chatbot with coding help. All the opposite main AI chatbots tout their coding capabilities. But Gemini Pro is the primary I’ve meaningfully examined for that function.)

    I requested Gemini to construct me a software that would scan my articles and add acceptable hyperlinks to different CNET articles. In my dialog with Gemini to construct and check the software, I defined in plain language any points, and Gemini would give you an answer and repair the code.

    It wasn’t an ideal expertise, although. There have been situations when Gemini would generate an up to date piece of code solely to go away out a characteristic that was within the prior model. I’d then ask Gemini the place that characteristic went, and it might apologize and repair the code. 

    Interestingly, once we’d hit roadblocks and I’d recommend that possibly the characteristic I used to be envisioning was just too troublesome to implement, Gemini would push again. It would say the characteristic was nonetheless completely doable and would generate revised code. This back-and-forth would go on till Gemini acquired it proper. 

    The bigger laptop science job market is presently going by means of an upheaval as Big Tech executives proceed shedding hundreds of staff whereas boasting about how a lot coding AI is doing. After utilizing Gemini to code, I perceive why college students are fearful. 

    Regardless, coding with Gemini has modified my understanding of the ability of AI chatbots, and I plan to vibe code extra as a part of the testing I do for critiques.

    Gemini Pro is (surprisingly) worse than ChatGPT for buying

    Searching for any product on Google Search results in an obnoxious mixture of product carousels and sponsored listings, drowning out all different critiques and articles (together with CNET’s). Considering how a lot Google invests in monetizing on-line buying, it is stunning how quaint the product analysis is on Gemini by comparability. ChatGPT has a much more strong buying expertise.

    Gemini is a wonderful software for primary product analysis. When I requested it to check varied fashions of Panasonic micro four-thirds cameras, for instance, Gemini might pull up fashions that met my specs and will tabulate their options in a helpful checklist when requested. It might add extra merchandise to that checklist as I continued to fall down the DSLR rabbit gap. 

    At the identical time, not like ChatGPT, it would not present hyperlinks to shops and it would not incorporate photographs. Product analysis on Gemini required me to have a separate Google Search window open simply so I might see present pricing and what varied digital camera fashions appeared like, facet by facet. Gemini can even typically get product particulars mistaken. But ChatGPT would additionally hyperlink to incorrect merchandise.

    Shopping is a kind of situations the place Gemini must act much less like an AI chatbot and extra like Google Search. I appeared for a chunk of furnishings to carry my file participant and retailer my vinyl, and Gemini gave me a information to what to search for when buying however did not truly suggest any merchandise. Jumping over to ChatGPT, it was a completely totally different expertise. There, it was like I used to be working with a gross sales affiliate at a furnishings retailer, going by means of the assorted choices to search out one thing that match my wants. 

    Image technology: At least it is higher than Gemini Free

    I did not extensively check picture technology with Gemini Pro, however I discovered it greater than satisfactory for primary duties. (CNET could have a separate hands-on of Google’s varied picture and video technology instruments.) At the very least, compared with Gemini Free, Gemini Pro did a greater job of following my intent when creating photographs. 

    Like with my Gemini Free overview, I needed to create a nostalgic picture that evoked the sensation of enjoying a Game Boy on a late night time drive within the again seat of a automobile. Gemini Pro acquired it on the primary go.
    Google Gemini Pro generates a picture of a boy enjoying his Game Boy within the backseat of a automobile at night time. Imad Khan/CNET
     My immediate that generated the picture above: “I want to create an image. One that evokes a feeling of nostalgia. The image is that of a boy playing his Game Boy in the back of his parents’ car on a long road trip at night. Because the screen isn’t backlit, he’s sitting near the window, trying to catch whatever passing light he can to see where to go. This image should use cool colors accented by the warmth of the light entering in. Feature anime-style artwork with a slight Western design. Should feel hand-drawn with intricately detailed linework. Analog VHS distortion. ’90s cartoon aesthetic.”
    My expertise with Gemini Free picture technology was rather more irritating. The mannequin merely did not perceive world logic and would usually place the boy within the entrance seat dealing with backward, or with surrounding autos driving within the mistaken route. Eventually, I gave up.

    Redemption

    Google’s accomplished it. After Bard’s dismal launch and a bumpy rebrand as Gemini, the newest construct of the corporate’s AI chatbot can compete with one of the best from OpenAI. This time round, Gemini brings with it higher accuracy, collaborative functionality, coding energy and picture technology to make an total compelling product.

    The Gemini 2.5 Pro mannequin is just higher than the free Gemini Flash mannequin. Answers have extra nuance and density, and options just like the picture generator work significantly higher. It’s a tad annoying when Gemini will default to the Flash mannequin even for Pro subscribers. I believe it does so when visitors is excessive and Google is making an attempt to minimize the load. It’s straightforward sufficient to change again, however it’s important to discover the change to take action.

    Compared with ChatGPT’s o3 mannequin, particularly, Gemini 2.5 Pro is quicker whereas sustaining comparable reply high quality. Google additionally says it has a 1 million token context window, which might dwarf what’s been reported concerning ChatGPT. Being capable of pull in information from different Google companies offers Gemini one other edge. 

    Gemini is not good, nonetheless. It can nonetheless stumble with some forms of queries, and utilizing it for buying is lackluster. Despite my qualms, I discovered myself more and more reliant on Gemini, transferring myself additional away from Google Search. 

    AI is slowly transferring Google away from being a search firm to a solution firm. Last yr, Gemini’s solutions have been too usually mistaken to be price recommending. This time, I’ve rather more confidence in Gemini’s solutions. Of course, if I ever publish something incorrect, the accountability will fall on me. 

    Ultimately, Gemini Pro acts as an expert and helpful know-it-all assistant. It would not have the angle of Claude or the controversy of Grok. Instead, it is there to assist even when you end up giving up. It’s that assertiveness that makes Gemini Pro a standout AI product.

    Recent Articles

    Related Stories

    Stay on op - Ge the daily news in your inbox