{ "text": "You've probably typed something into Gemini, gotten an answer, and closed the tab.\nThe same way you'd use any other chatbot. Here's the thing. That's maybe 10% of what it actually\ndoes. Use it right, and it can quietly take over almost half the busy work you're still doing by\nhand. I spent hours mapping every model, every mode, and every product Gemini is quietly wired\ninto. And the number that stopped me was this. Gemini's app alone has over 650 million monthly\nusers and that's before you count everyone using it inside search. Most of them are using maybe 20%\nof it with no idea the rest even exists. Look at this data from the Census Bureau.\nOnly about one in five U.S. businesses actually use AI in their operations.\nSo if you run a business and you're even thinking about this, you're ahead of most of your\ncompetition. What you might not know is that alongside covering AI news, we work with business\nowners to help them implement AI in their business. Our engineering team gets to know\nhow your business runs, then builds the automation with you. You'll find the link in the description\nbelow. Click it, fill out a short form about your business, and we'll get in touch to set up a call.\nSo in this video, I'm breaking down exactly what Gemini is as of mid-2026. Every current model,\nevery mode, and everywhere Google has quietly built it in. By the end, you'll know exactly\nwhich Gemini tool to reach for depending on what you're actually trying to do, instead of just\ntyping into whichever box is in front of you. First, let's clear up the biggest misconception.\nGemini isn't one product at all. What Gemini actually is. Here's the mental model you need\nbefore any of this makes sense. Gemini isn't a single AI. It's Google's umbrella name for a\nwhole platform, a family of models underneath, and a set of products on top that let you actually\ntalk to them. Think of it in two layers. The bottom layer is the models themselves. Things\nlike Gemini 3.6 Flash or Gemini 3.1 Pro. These are the engines, tuned for different jobs. Some\nbuilt for speed, some for heavy reasoning, some for images or audio. You never see these names\nunless you go looking. The top layer is everything you actually click on. The Gemini app, AI mode\ninside Google search, Gemini inside Gmail and Docs, the voice assistant on your phone. All of those\nare just different doors into the same underlying models. That's the whole point of this video.\nGoogle isn't trying to build one great chatbot. It's trying to put the same AI brain behind every\nproduct you already use. So let's start with the brains, the actual models, because once you know\nwhat each one is built for, everything else clicks into place. The current model lineup.\nThis is a demo checklist, so we're going model by model. What it is, what it's actually good for,\nand where you can get it. Gemini 3.7 Flash. Launched on August 13, 2026, this is Google's\nnewest Flash model and its most capable workhorse yet. It's built primarily for coding and AI agents,\nwith major improvements in software engineering, web development, and complex multi-step workflows.\nGoogle has already made it generally available through the Gemini API, positioning 3.7 Flash\nas the new go-to model when you want strong intelligence without giving up the speed and\nefficiency the Flash lineup is known for. Gemini 3.6 Flash. This is Google's current flagship,\nannounced in a company blog post on July 21, 2026. It's built as a workhorse, strong at coding,\nknowledge work, and multimodal tasks. And according to Google's own numbers,\nit does the job using about 17% fewer tokens on average than its predecessor.\nFewer tokens means faster answers and a lower bill if you're paying for it through the API.\nYou can reach it through the Gemini API, through AI Studio, or simply by using the Gemini app and Search's AI mode.\nNo extra setup required.\nIf you only remember one model name from this video, make it this one, because it's what most of Gemini is quietly running on right now.\nGemini 3.5 Flash.\nThis one launched back in May 2026, and it's the model that was actually powering AI mode in Search before 3.6 Flash took over.\nGoogle described it as frontier-level intelligence at exceptional speed,\nand by its own benchmarks, it pushed output throughput to roughly four times faster than\nother top models at the time. It's still very much in active use across the Gemini app,\nGoogle's anti-gravity platform, and enterprise tools. It's slightly behind 3.6 now, but it's\nthe model quietly sitting behind a huge share of what shipped this year, Gemini 3.5 flashlight.\nSame July announcement, different job entirely.\nThis is the stripped-down, high-throughput sibling Google cites roughly 350 tokens per second,\nwhich is built for volume, not depth.\nYou wouldn't use this for a hard reasoning problem.\nYou'd use it for background tasks and agent pipelines that need to move fast and cheap at scale.\nGemini 3.1 Pro\nReleased in February 2026, this is the reasoning specialist.\nGoogle's benchmarks claim roughly double the logic performance of the earlier Gemini 3 Pro.\nIt's mostly gated behind preview access through the API, Antigravity, Vertex AI, and Pro or Ultra subscriptions in the consumer app.\nIf 3.6 Flash is built for speed, 3.1 Pro is built for depth.\nThe model you'd want on a genuinely hard problem, not a quick one.\nAnd if you're actually building on top of these through the API, price is where the real-world decision gets made.\nAccording to Google's own published rates, 3.6 Flash runs about $1.50 per million input tokens\nand 750 per million output tokens.\nFor context, that's noticeably cheaper on the output side\nthan GPT 5.6 LUNA's roughly $6 per million.\nThat gap is exactly why so many developers default to flash tier models\nfor anything running at volume.\nNow, a handful of specialty models worth knowing by name,\neven if we don't dwell on each one.\nNano Banana 2 is Gemini's current image generation and editing model,\nreplacing the older Imogen line entirely,\nAnd that's not a small detail, because Amagen is actually shutting down on August 17,\n2026. If you've had workflows built on Amagen, that clock is already running. There's also a\nNano Banana 2 Lite variant built purely for speed, trading a small amount of quality for\nmuch faster, cheaper output at high volume. VO 3.1 is Google's video generation model,\nstill in beta, built to turn a text prompt into a short clip with matching audio.\nGemini Audio 3.5 Live Translate handles real-time speech-to-speech translation\nacross more than 70 languages already built into Google Meet and Android.\nAnd if you're curious about the more niche end of the lineup,\nthere's a security-focused variant called 3.5 Flash Cyber Built\nto coordinate with vulnerability scanning tools,\nand Lyria 3.5, Google's music model,\nwhich can now generate tracks up to three minutes long from a text prompt.\nHere's the honest limitation worth naming.\nGoogle ships a lot of these models fast, and the naming gets confusing on purpose or not.\n3.5, 3.6, 3.1 Pro, Flashlight, FlashCyber.\nIf you're not building on top of the API professionally, you genuinely don't need to memorize this list.\nYou just need to know the shape of it. Fast and cheap, deep reasoning, and multimodal.\nThat's really three categories wearing a lot of different name tags.\nThe modes you actually interact with models are the engine.\nModes are the steering wheel. Here's where things get useful for anyone who isn't a developer.\nAI mode inside Google search turns your search bar into a conversation. As of IO 2026,\nit's globally powered by Gemini 3.5 Flash, and instead of 10 blue links, you get a written\nanswer with follow-up questions and sometimes an interactive widget built on the fly. Anyone\nwith search can use it. No subscription required. Ask something like, what's a quick dinner with\nwhat's in my fridge and it answers in full sentences not a list of recipes\nDeepThink is the extra effort version of the Gemini app.\nIt spends more compute per answer to reason through harder problems step by step.\nGoogle gates this one behind Google AI Ultra,\nand it's built for genuinely difficult science or engineering questions, not everyday chat.\nNow, deep research is where this stops being a chatbot and starts being an assistant.\nYou give it a topic, and instead of one reply, it plans a research strategy,\nopens webpages, reads them, and, if you allow it, pulls from your own Gmail and Drive too.\nWhat comes back isn't a paragraph.\nIt's a full multi-page report inside Gemini's Canvas.\nThis is the part of Gemini that actually earns the word agent, and we're coming back to why that matters in a few minutes.\nGemini Live is the voice and camera mode.\nSay, hey Google, let's chat, and you're talking to it hands-free, with the option to point your camera at something and ask what it's looking at.\nLive.\nAnd Canvas is the workspace mode.\nType, create a quiz app about planets, and it writes the code, the interface, and the content in one pass.\nRight there for you to edit.\nOne more worth a mention, briefly, because it's still early.\nGemini Spark, a personal agent announced at IO 2026,\nmeant to run continuously in the background handling things like scheduling.\nRight now it's limited to early ultra testers.\nSo treat this one as coming, not here.\nQuick gut check before we move on.\nIf all of that sounds like a lot of separate tools, that's fair.\nBut notice the pattern.\nEvery single one of these modes is just Gemini 3.5 or 3.6 Flash wearing a different job title.\nYou're not learning six different AIs.\nYou're learning six different ways to ask the same brain for help.\nMultimodal and practice.\nLet's talk about what multimodal actually means day to day,\nbecause it's more than a buzzword on a slide.\nGemini reads and writes text and code.\nObviously, that's the baseline.\nBut drop a photo into a chat and ask it to caption or edit it,\nand Nano Banana handles that.\nAsk it to speak and answer out loud,\nand Gemini's audio models generate that voice on the spot\nwith actual control over tone and pacing.\nAsk for a short video, and VO builds one from scratch.\nAsk it to edit an existing clip.\nSwap the sky, change the style, and that's a separate tool called Gemini Omni,\ndoing frame-by-frame editing by voice command.\nInside Google Docs and Sheets, the same underlying models can draft a document from your meeting notes\nor build a spreadsheet out of a pile of invoices, complete with formulas and charts,\nnot just raw numbers dumped into cells.\nIn Slides, hand it a list of bullet points and it can lay out an actual deck, not just text on blank slides.\nAnd through Gemini Live's camera mode, you can point your phone at a menu in a language you don't speak\nand get a live translation overlaid on what you're looking at,\nor ask it to identify an object it's looking at through the lens.\nNo typing involved.\nHere's the part worth remembering.\nYou never pick the model.\nYou just say what you want.\nMake this an infographic.\nTranslate this.\nWrite this in Python.\nand Gemini quietly roots the request to whichever model actually does that job.\nThat's the design philosophy in one sentence.\nOne platform, and it decides the plumbing so you don't have to.\nWhere Gemini actually lives.\nThis is the part that's easy to underestimate.\nGemini isn't confined to one app.\nIt's spread across nearly everything Google ships.\nIn search, it's AI mode, already covered.\nIn Gmail, it's behind Smart Compose and Auto Reply suggestions.\nIn Docs, Sheets and Slides.\nUltra and Pro subscribers get Gemini drafting text, building formulas, and designing slide layouts,\npulling context from your own files when you let it.\nIn Drive, it can find and summarize documents for you.\nIn Google Meet, it's doing live caption translation.\nOn Android, especially Pixel devices, it's baked straight into the Voice Assistant,\nand there's a Chrome extension that lets the browser send page content straight to Gemini\nso you can ask questions about whatever tab you're on.\nAnd for developers, all of it is exposed through Google AI Studio and the Gemini API,\nplus a newer platform called Anti-Gravity for building multi-agent workflows on top of it.\nThe strategic point here isn't subtle.\nGoogle isn't trying to win the best standalone chatbot argument.\nIt's trying to make sure you're never more than one product away from Gemini,\nno matter what you're doing on a Google device or in a Google app.\nAgents, the part that actually matters.\nNow here's the shift I promised earlier.\nthe one that actually changes what this platform is for.\nEverything so far has been ask a question, get an answer.\nAgents are Google trying to move Gemini past that entirely.\nDeep research is the clearest example already live, plan, browse, synthesize, write,\nwithout you babysitting every step.\nSpark is the early, still limited attempt at a persistent personal agent\nrunning continuously in the background.\nAnd on the developer side,\nanti-gravity lets companies build coordinated teams of sub-agents. Google's own blog post gave\nan example of businesses running parallel agents to analyze data at scale, rather than one model\ndoing everything sequentially. Picture the difference in practice. The old way. You ask\nGemini, what should I know before a trip to Japan? And it gives you a paragraph. The agent way. You\nsay, plan my trip to Japan. And it checks flights, compares hotel options, and comes back with an\nactual itinerary, pausing to confirm with you before it books anything. That's the same underlying\nmodel, just given permission to take more than one step before handing control back to you.\nNone of this is science fiction anymore, and none of it is fully finished either.\nThat's the honest read. Deep research genuinely works today. Spark is still in early testing,\nbut the direction is unmistakable. Google wants Gemini to eventually take a task,\nbreak it into steps, and execute most of them without you typing a follow-up for every single\none. What actually makes Gemini different? So how does this stack up against everyone else building\nthe same kind of thing? Let's be balanced here, because Google's advantages are real, but so are\nits weak spots. The clearest edge is data. Gemini can pull from live search results, maps, Gmail,\nand Drive in ways that a closed, sandbox chatbot simply can't match without plugins bolted on.\nThe second edge is reach. Every Android phone is a potential Gemini client, and every workspace\nbusiness account already has it available. No competitor has that kind of built-in distribution.\nAnd on raw benchmarks, Gemini 3 Pro topped the LM Arena leaderboard, which, regardless of how much\nweight you put on any single leaderboard, says Google's infrastructure and DeepMind's research\nare producing real, top-tier results, not just hype. But, and this matters for credibility,\nGoogle is genuinely more conservative about rollout than some competitors.\nDeepThink and Spark are still gated behind ultra subscriptions or limited testing, while\nsome rivals ship new capabilities to everyone at once.\nAnd the tier structure itself — free, pro, ultra, API pricing — can be genuinely confusing\nnext to a simpler flat subscription from a competitor.\nIf you've ever opened the Gemini pricing page and closed it five minutes later still\nunsure which plan you need, that's not just you.\nWhere this is actually headed.\nA few things are confirmed, and a few are still rumor, and it's worth keeping those separate.\nConfirmed. Gemini 3.5 Pro is currently in partner testing with a public release expected soon,\nand Google has already started training on Gemini 4, according to its own July 2026 announcement,\nthough there's no public timeline for that yet. Workspace AI rollout continues expanding,\nand Gemini Live's regional language support keeps growing. Speculative, and worth labeling clearly\nas such. There's talk of a dedicated on-device AI chip for future Pixel phones, and some experimental\ndeep-mind research around 3D avatars and world simulation that hasn't shipped as a product.\nTreat both of those as possible, not coming. Nothing official has confirmed either one.\nThe verdict. So where does that leave things? Gemini in 2026 isn't a chatbot you occasionally\nopen. It's an AI layer Google has threaded through search, Gmail, your documents, and increasingly,\nyour phone itself. The models handle the thinking, the modes handle how you ask,\nand agents like Deep Research are the clearest sign of where all of it is actually heading.\nIf there's one thing worth trying this week, it's Deep Research on something you'd normally\nspend an evening looking into yourself, and actually watching it work instead of just\nreading the final report. Drop a comment with which piece of this surprised you most,\nthe model lineup, the agent side, or just how much of this you were already using without\nrealizing it. I'll be back soon with a deeper breakdown on how deep research actually performs\nagainst a real research task. Thanks for watching, and I'll see you in the next one.", "language": "en", "duration": 997.248, "segments": [ { "start": 0.0, "end": 3.98, "text": "You've probably typed something into Gemini, gotten an answer, and closed the tab.", "chunk": 0, "language": "en", "avg_logprob": -0.08565459817142809, "no_speech_prob": 2.9167731598056523e-12, "compression_ratio": 1.595890410958904 }, { "start": 4.5, "end": 9.72, "text": "The same way you'd use any other chatbot. Here's the thing. That's maybe 10% of what it actually", "chunk": 0, "language": "en", "avg_logprob": -0.08565459817142809, "no_speech_prob": 2.9167731598056523e-12, "compression_ratio": 1.595890410958904 }, { "start": 9.72, "end": 15.12, "text": "does. Use it right, and it can quietly take over almost half the busy work you're still doing by", "chunk": 0, "language": "en", "avg_logprob": -0.08565459817142809, "no_speech_prob": 2.9167731598056523e-12, "compression_ratio": 1.595890410958904 }, { "start": 15.12, "end": 21.36, "text": "hand. I spent hours mapping every model, every mode, and every product Gemini is quietly wired", "chunk": 0, "language": "en", "avg_logprob": -0.08565459817142809, "no_speech_prob": 2.9167731598056523e-12, "compression_ratio": 1.595890410958904 }, { "start": 21.36, "end": 27.48, "text": "into. And the number that stopped me was this. Gemini's app alone has over 650 million monthly", "chunk": 0, "language": "en", "avg_logprob": -0.08565459817142809, "no_speech_prob": 2.9167731598056523e-12, "compression_ratio": 1.595890410958904 }, { "start": 27.48, "end": 33.64, "text": "users and that's before you count everyone using it inside search. Most of them are using maybe 20%", "chunk": 0, "language": "en", "avg_logprob": -0.09422876216747143, "no_speech_prob": 2.201837236842641e-12, "compression_ratio": 1.608695652173913 }, { "start": 33.64, "end": 38.04, "text": "of it with no idea the rest even exists. Look at this data from the Census Bureau.", "chunk": 0, "language": "en", "avg_logprob": -0.09422876216747143, "no_speech_prob": 2.201837236842641e-12, "compression_ratio": 1.608695652173913 }, { "start": 39.0, "end": 43.32, "text": "Only about one in five U.S. businesses actually use AI in their operations.", "chunk": 0, "language": "en", "avg_logprob": -0.09422876216747143, "no_speech_prob": 2.201837236842641e-12, "compression_ratio": 1.608695652173913 }, { "start": 44.36, "end": 48.6, "text": "So if you run a business and you're even thinking about this, you're ahead of most of your", "chunk": 0, "language": "en", "avg_logprob": -0.09422876216747143, "no_speech_prob": 2.201837236842641e-12, "compression_ratio": 1.608695652173913 }, { "start": 48.6, "end": 54.68, "text": "competition. What you might not know is that alongside covering AI news, we work with business", "chunk": 0, "language": "en", "avg_logprob": -0.09422876216747143, "no_speech_prob": 2.201837236842641e-12, "compression_ratio": 1.608695652173913 }, { "start": 54.68, "end": 59.32, "text": "owners to help them implement AI in their business. Our engineering team gets to know", "chunk": 0, "language": "en", "avg_logprob": -0.04979276657104492, "no_speech_prob": 3.2536217643708287e-12, "compression_ratio": 1.6067796610169491 }, { "start": 59.32, "end": 64.44, "text": "how your business runs, then builds the automation with you. You'll find the link in the description", "chunk": 0, "language": "en", "avg_logprob": -0.04979276657104492, "no_speech_prob": 3.2536217643708287e-12, "compression_ratio": 1.6067796610169491 }, { "start": 64.44, "end": 70.04, "text": "below. Click it, fill out a short form about your business, and we'll get in touch to set up a call.", "chunk": 0, "language": "en", "avg_logprob": -0.04979276657104492, "no_speech_prob": 3.2536217643708287e-12, "compression_ratio": 1.6067796610169491 }, { "start": 70.04, "end": 77.64, "text": "So in this video, I'm breaking down exactly what Gemini is as of mid-2026. Every current model,", "chunk": 0, "language": "en", "avg_logprob": -0.04979276657104492, "no_speech_prob": 3.2536217643708287e-12, "compression_ratio": 1.6067796610169491 }, { "start": 77.64, "end": 83.24, "text": "every mode, and everywhere Google has quietly built it in. By the end, you'll know exactly", "chunk": 0, "language": "en", "avg_logprob": -0.04979276657104492, "no_speech_prob": 3.2536217643708287e-12, "compression_ratio": 1.6067796610169491 }, { "start": 83.24, "end": 88.16, "text": "which Gemini tool to reach for depending on what you're actually trying to do, instead of just", "chunk": 0, "language": "en", "avg_logprob": -0.04090044233534071, "no_speech_prob": 1.579590516546936e-12, "compression_ratio": 1.6491228070175439 }, { "start": 88.16, "end": 92.76, "text": "typing into whichever box is in front of you. First, let's clear up the biggest misconception.", "chunk": 0, "language": "en", "avg_logprob": -0.04090044233534071, "no_speech_prob": 1.579590516546936e-12, "compression_ratio": 1.6491228070175439 }, { "start": 93.38, "end": 98.94, "text": "Gemini isn't one product at all. What Gemini actually is. Here's the mental model you need", "chunk": 0, "language": "en", "avg_logprob": -0.04090044233534071, "no_speech_prob": 1.579590516546936e-12, "compression_ratio": 1.6491228070175439 }, { "start": 98.94, "end": 104.3, "text": "before any of this makes sense. Gemini isn't a single AI. It's Google's umbrella name for a", "chunk": 0, "language": "en", "avg_logprob": -0.04090044233534071, "no_speech_prob": 1.579590516546936e-12, "compression_ratio": 1.6491228070175439 }, { "start": 104.3, "end": 109.4, "text": "whole platform, a family of models underneath, and a set of products on top that let you actually", "chunk": 0, "language": "en", "avg_logprob": -0.04090044233534071, "no_speech_prob": 1.579590516546936e-12, "compression_ratio": 1.6491228070175439 }, { "start": 109.4, "end": 114.22, "text": "talk to them. Think of it in two layers. The bottom layer is the models themselves. Things", "chunk": 0, "language": "en", "avg_logprob": -0.03409581887917441, "no_speech_prob": 4.064884279869618e-12, "compression_ratio": 1.6426116838487972 }, { "start": 114.22, "end": 121.18, "text": "like Gemini 3.6 Flash or Gemini 3.1 Pro. These are the engines, tuned for different jobs. Some", "chunk": 0, "language": "en", "avg_logprob": -0.03409581887917441, "no_speech_prob": 4.064884279869618e-12, "compression_ratio": 1.6426116838487972 }, { "start": 121.18, "end": 126.64, "text": "built for speed, some for heavy reasoning, some for images or audio. You never see these names", "chunk": 0, "language": "en", "avg_logprob": -0.03409581887917441, "no_speech_prob": 4.064884279869618e-12, "compression_ratio": 1.6426116838487972 }, { "start": 126.64, "end": 132.32, "text": "unless you go looking. The top layer is everything you actually click on. The Gemini app, AI mode", "chunk": 0, "language": "en", "avg_logprob": -0.03409581887917441, "no_speech_prob": 4.064884279869618e-12, "compression_ratio": 1.6426116838487972 }, { "start": 132.32, "end": 138.4, "text": "inside Google search, Gemini inside Gmail and Docs, the voice assistant on your phone. All of those", "chunk": 0, "language": "en", "avg_logprob": -0.03409581887917441, "no_speech_prob": 4.064884279869618e-12, "compression_ratio": 1.6426116838487972 }, { "start": 138.4, "end": 142.8, "text": "are just different doors into the same underlying models. That's the whole point of this video.", "chunk": 0, "language": "en", "avg_logprob": -0.09639995933597924, "no_speech_prob": 3.67274613499724e-12, "compression_ratio": 1.6771929824561405 }, { "start": 143.2, "end": 148.3, "text": "Google isn't trying to build one great chatbot. It's trying to put the same AI brain behind every", "chunk": 0, "language": "en", "avg_logprob": -0.09639995933597924, "no_speech_prob": 3.67274613499724e-12, "compression_ratio": 1.6771929824561405 }, { "start": 148.3, "end": 154.12, "text": "product you already use. So let's start with the brains, the actual models, because once you know", "chunk": 0, "language": "en", "avg_logprob": -0.09639995933597924, "no_speech_prob": 3.67274613499724e-12, "compression_ratio": 1.6771929824561405 }, { "start": 154.12, "end": 158.88, "text": "what each one is built for, everything else clicks into place. The current model lineup.", "chunk": 0, "language": "en", "avg_logprob": -0.09639995933597924, "no_speech_prob": 3.67274613499724e-12, "compression_ratio": 1.6771929824561405 }, { "start": 159.26, "end": 164.88, "text": "This is a demo checklist, so we're going model by model. What it is, what it's actually good for,", "chunk": 0, "language": "en", "avg_logprob": -0.09639995933597924, "no_speech_prob": 3.67274613499724e-12, "compression_ratio": 1.6771929824561405 }, { "start": 164.88, "end": 172.02, "text": "and where you can get it. Gemini 3.7 Flash. Launched on August 13, 2026, this is Google's", "chunk": 0, "language": "en", "avg_logprob": -0.10411097096130911, "no_speech_prob": 2.5132253049509057e-12, "compression_ratio": 1.5457516339869282 }, { "start": 172.02, "end": 177.56, "text": "newest Flash model and its most capable workhorse yet. It's built primarily for coding and AI agents,", "chunk": 0, "language": "en", "avg_logprob": -0.10411097096130911, "no_speech_prob": 2.5132253049509057e-12, "compression_ratio": 1.5457516339869282 }, { "start": 177.82, "end": 183.22, "text": "with major improvements in software engineering, web development, and complex multi-step workflows.", "chunk": 0, "language": "en", "avg_logprob": -0.10411097096130911, "no_speech_prob": 2.5132253049509057e-12, "compression_ratio": 1.5457516339869282 }, { "start": 183.94, "end": 189.6, "text": "Google has already made it generally available through the Gemini API, positioning 3.7 Flash", "chunk": 0, "language": "en", "avg_logprob": -0.10411097096130911, "no_speech_prob": 2.5132253049509057e-12, "compression_ratio": 1.5457516339869282 }, { "start": 189.6, "end": 194.1, "text": "as the new go-to model when you want strong intelligence without giving up the speed and", "chunk": 0, "language": "en", "avg_logprob": -0.10411097096130911, "no_speech_prob": 2.5132253049509057e-12, "compression_ratio": 1.5457516339869282 }, { "start": 194.1, "end": 199.88, "text": "efficiency the Flash lineup is known for. Gemini 3.6 Flash. This is Google's current flagship,", "chunk": 0, "language": "en", "avg_logprob": -0.10105233533041817, "no_speech_prob": 3.2282343031403427e-12, "compression_ratio": 1.5017182130584192 }, { "start": 200.42, "end": 206.62, "text": "announced in a company blog post on July 21, 2026. It's built as a workhorse, strong at coding,", "chunk": 0, "language": "en", "avg_logprob": -0.10105233533041817, "no_speech_prob": 3.2282343031403427e-12, "compression_ratio": 1.5017182130584192 }, { "start": 207.02, "end": 210.92, "text": "knowledge work, and multimodal tasks. And according to Google's own numbers,", "chunk": 0, "language": "en", "avg_logprob": -0.10105233533041817, "no_speech_prob": 3.2282343031403427e-12, "compression_ratio": 1.5017182130584192 }, { "start": 211.36, "end": 215.96, "text": "it does the job using about 17% fewer tokens on average than its predecessor.", "chunk": 0, "language": "en", "avg_logprob": -0.10105233533041817, "no_speech_prob": 3.2282343031403427e-12, "compression_ratio": 1.5017182130584192 }, { "start": 216.36, "end": 221.14, "text": "Fewer tokens means faster answers and a lower bill if you're paying for it through the API.", "chunk": 0, "language": "en", "avg_logprob": -0.10105233533041817, "no_speech_prob": 3.2282343031403427e-12, "compression_ratio": 1.5017182130584192 }, { "start": 221.14, "end": 229.08, "text": "You can reach it through the Gemini API, through AI Studio, or simply by using the Gemini app and Search's AI mode.", "chunk": 0, "language": "en", "avg_logprob": -0.08870283762613933, "no_speech_prob": 1.6873839565195725e-12, "compression_ratio": 1.5608856088560885 }, { "start": 229.44, "end": 230.82, "text": "No extra setup required.", "chunk": 0, "language": "en", "avg_logprob": -0.08870283762613933, "no_speech_prob": 1.6873839565195725e-12, "compression_ratio": 1.5608856088560885 }, { "start": 231.18, "end": 237.8, "text": "If you only remember one model name from this video, make it this one, because it's what most of Gemini is quietly running on right now.", "chunk": 0, "language": "en", "avg_logprob": -0.08870283762613933, "no_speech_prob": 1.6873839565195725e-12, "compression_ratio": 1.5608856088560885 }, { "start": 238.12, "end": 239.8, "text": "Gemini 3.5 Flash.", "chunk": 0, "language": "en", "avg_logprob": -0.08870283762613933, "no_speech_prob": 1.6873839565195725e-12, "compression_ratio": 1.5608856088560885 }, { "start": 240.12, "end": 248.82, "text": "This one launched back in May 2026, and it's the model that was actually powering AI mode in Search before 3.6 Flash took over.", "chunk": 0, "language": "en", "avg_logprob": -0.08870283762613933, "no_speech_prob": 1.6873839565195725e-12, "compression_ratio": 1.5608856088560885 }, { "start": 248.82, "end": 252.94, "text": "Google described it as frontier-level intelligence at exceptional speed,", "chunk": 0, "language": "en", "avg_logprob": -0.08065497200444059, "no_speech_prob": 8.620715361180864e-13, "compression_ratio": 1.6043956043956045 }, { "start": 253.3, "end": 258.1, "text": "and by its own benchmarks, it pushed output throughput to roughly four times faster than", "chunk": 0, "language": "en", "avg_logprob": -0.08065497200444059, "no_speech_prob": 8.620715361180864e-13, "compression_ratio": 1.6043956043956045 }, { "start": 258.1, "end": 262.74, "text": "other top models at the time. It's still very much in active use across the Gemini app,", "chunk": 0, "language": "en", "avg_logprob": -0.08065497200444059, "no_speech_prob": 8.620715361180864e-13, "compression_ratio": 1.6043956043956045 }, { "start": 263.22, "end": 269.22, "text": "Google's anti-gravity platform, and enterprise tools. It's slightly behind 3.6 now, but it's", "chunk": 0, "language": "en", "avg_logprob": -0.08065497200444059, "no_speech_prob": 8.620715361180864e-13, "compression_ratio": 1.6043956043956045 }, { "start": 269.22, "end": 274.84, "text": "the model quietly sitting behind a huge share of what shipped this year, Gemini 3.5 flashlight.", "chunk": 0, "language": "en", "avg_logprob": -0.08065497200444059, "no_speech_prob": 8.620715361180864e-13, "compression_ratio": 1.6043956043956045 }, { "start": 274.84, "end": 277.78, "text": "Same July announcement, different job entirely.", "chunk": 0, "language": "en", "avg_logprob": -0.1407373146932633, "no_speech_prob": 1.1115198388436665e-12, "compression_ratio": 1.5434782608695652 }, { "start": 278.24, "end": 284.02, "text": "This is the stripped-down, high-throughput sibling Google cites roughly 350 tokens per second,", "chunk": 0, "language": "en", "avg_logprob": -0.1407373146932633, "no_speech_prob": 1.1115198388436665e-12, "compression_ratio": 1.5434782608695652 }, { "start": 284.38, "end": 286.32, "text": "which is built for volume, not depth.", "chunk": 0, "language": "en", "avg_logprob": -0.1407373146932633, "no_speech_prob": 1.1115198388436665e-12, "compression_ratio": 1.5434782608695652 }, { "start": 286.62, "end": 288.54, "text": "You wouldn't use this for a hard reasoning problem.", "chunk": 0, "language": "en", "avg_logprob": -0.1407373146932633, "no_speech_prob": 1.1115198388436665e-12, "compression_ratio": 1.5434782608695652 }, { "start": 288.86, "end": 293.98, "text": "You'd use it for background tasks and agent pipelines that need to move fast and cheap at scale.", "chunk": 0, "language": "en", "avg_logprob": -0.1407373146932633, "no_speech_prob": 1.1115198388436665e-12, "compression_ratio": 1.5434782608695652 }, { "start": 294.34, "end": 295.74, "text": "Gemini 3.1 Pro", "chunk": 0, "language": "en", "avg_logprob": -0.1407373146932633, "no_speech_prob": 1.1115198388436665e-12, "compression_ratio": 1.5434782608695652 }, { "start": 295.74, "end": 299.8, "text": "Released in February 2026, this is the reasoning specialist.", "chunk": 0, "language": "en", "avg_logprob": -0.1407373146932633, "no_speech_prob": 1.1115198388436665e-12, "compression_ratio": 1.5434782608695652 }, { "start": 299.96, "end": 304.7, "text": "Google's benchmarks claim roughly double the logic performance of the earlier Gemini 3 Pro.", "chunk": 0, "language": "en", "avg_logprob": -0.1407373146932633, "no_speech_prob": 1.1115198388436665e-12, "compression_ratio": 1.5434782608695652 }, { "start": 304.84, "end": 313.6, "text": "It's mostly gated behind preview access through the API, Antigravity, Vertex AI, and Pro or Ultra subscriptions in the consumer app.", "chunk": 0, "language": "en", "avg_logprob": -0.14698013305664062, "no_speech_prob": 1.4775336986891374e-12, "compression_ratio": 1.540983606557377 }, { "start": 313.6, "end": 318.48, "text": "If 3.6 Flash is built for speed, 3.1 Pro is built for depth.", "chunk": 0, "language": "en", "avg_logprob": -0.14698013305664062, "no_speech_prob": 1.4775336986891374e-12, "compression_ratio": 1.540983606557377 }, { "start": 318.48, "end": 322.2, "text": "The model you'd want on a genuinely hard problem, not a quick one.", "chunk": 0, "language": "en", "avg_logprob": -0.14698013305664062, "no_speech_prob": 1.4775336986891374e-12, "compression_ratio": 1.540983606557377 }, { "start": 322.2, "end": 327.88, "text": "And if you're actually building on top of these through the API, price is where the real-world decision gets made.", "chunk": 0, "language": "en", "avg_logprob": -0.14698013305664062, "no_speech_prob": 1.4775336986891374e-12, "compression_ratio": 1.540983606557377 }, { "start": 327.88, "end": 334.44, "text": "According to Google's own published rates, 3.6 Flash runs about $1.50 per million input tokens", "chunk": 0, "language": "en", "avg_logprob": -0.14698013305664062, "no_speech_prob": 1.4775336986891374e-12, "compression_ratio": 1.540983606557377 }, { "start": 334.44, "end": 337.04, "text": "and 750 per million output tokens.", "chunk": 0, "language": "en", "avg_logprob": -0.17878582517979508, "no_speech_prob": 9.693864634618965e-13, "compression_ratio": 1.5170068027210883 }, { "start": 337.04, "end": 340.64, "text": "For context, that's noticeably cheaper on the output side", "chunk": 0, "language": "en", "avg_logprob": -0.17878582517979508, "no_speech_prob": 9.693864634618965e-13, "compression_ratio": 1.5170068027210883 }, { "start": 340.64, "end": 344.84, "text": "than GPT 5.6 LUNA's roughly $6 per million.", "chunk": 0, "language": "en", "avg_logprob": -0.17878582517979508, "no_speech_prob": 9.693864634618965e-13, "compression_ratio": 1.5170068027210883 }, { "start": 344.84, "end": 349.34, "text": "That gap is exactly why so many developers default to flash tier models", "chunk": 0, "language": "en", "avg_logprob": -0.17878582517979508, "no_speech_prob": 9.693864634618965e-13, "compression_ratio": 1.5170068027210883 }, { "start": 349.34, "end": 351.04, "text": "for anything running at volume.", "chunk": 0, "language": "en", "avg_logprob": -0.17878582517979508, "no_speech_prob": 9.693864634618965e-13, "compression_ratio": 1.5170068027210883 }, { "start": 351.04, "end": 354.64, "text": "Now, a handful of specialty models worth knowing by name,", "chunk": 0, "language": "en", "avg_logprob": -0.17878582517979508, "no_speech_prob": 9.693864634618965e-13, "compression_ratio": 1.5170068027210883 }, { "start": 354.64, "end": 356.64, "text": "even if we don't dwell on each one.", "chunk": 0, "language": "en", "avg_logprob": -0.17878582517979508, "no_speech_prob": 9.693864634618965e-13, "compression_ratio": 1.5170068027210883 }, { "start": 356.64, "end": 361.24, "text": "Nano Banana 2 is Gemini's current image generation and editing model,", "chunk": 0, "language": "en", "avg_logprob": -0.17878582517979508, "no_speech_prob": 9.693864634618965e-13, "compression_ratio": 1.5170068027210883 }, { "start": 361.24, "end": 364.04, "text": "replacing the older Imogen line entirely,", "chunk": 0, "language": "en", "avg_logprob": -0.17878582517979508, "no_speech_prob": 9.693864634618965e-13, "compression_ratio": 1.5170068027210883 }, { "start": 364.04, "end": 368.84, "text": "And that's not a small detail, because Amagen is actually shutting down on August 17,", "chunk": 0, "language": "en", "avg_logprob": -0.10456640785987224, "no_speech_prob": 1.681202594673581e-12, "compression_ratio": 1.5034482758620689 }, { "start": 369.16, "end": 374.84, "text": "2026. If you've had workflows built on Amagen, that clock is already running. There's also a", "chunk": 0, "language": "en", "avg_logprob": -0.10456640785987224, "no_speech_prob": 1.681202594673581e-12, "compression_ratio": 1.5034482758620689 }, { "start": 374.84, "end": 380.04, "text": "Nano Banana 2 Lite variant built purely for speed, trading a small amount of quality for", "chunk": 0, "language": "en", "avg_logprob": -0.10456640785987224, "no_speech_prob": 1.681202594673581e-12, "compression_ratio": 1.5034482758620689 }, { "start": 380.04, "end": 385.72, "text": "much faster, cheaper output at high volume. VO 3.1 is Google's video generation model,", "chunk": 0, "language": "en", "avg_logprob": -0.10456640785987224, "no_speech_prob": 1.681202594673581e-12, "compression_ratio": 1.5034482758620689 }, { "start": 385.72, "end": 390.36, "text": "still in beta, built to turn a text prompt into a short clip with matching audio.", "chunk": 0, "language": "en", "avg_logprob": -0.10456640785987224, "no_speech_prob": 1.681202594673581e-12, "compression_ratio": 1.5034482758620689 }, { "start": 390.36, "end": 396.36, "text": "Gemini Audio 3.5 Live Translate handles real-time speech-to-speech translation", "chunk": 0, "language": "en", "avg_logprob": -0.1060038169225057, "no_speech_prob": 1.0315117900464021e-12, "compression_ratio": 1.544871794871795 }, { "start": 396.36, "end": 400.76, "text": "across more than 70 languages already built into Google Meet and Android.", "chunk": 0, "language": "en", "avg_logprob": -0.1060038169225057, "no_speech_prob": 1.0315117900464021e-12, "compression_ratio": 1.544871794871795 }, { "start": 400.76, "end": 403.56, "text": "And if you're curious about the more niche end of the lineup,", "chunk": 0, "language": "en", "avg_logprob": -0.1060038169225057, "no_speech_prob": 1.0315117900464021e-12, "compression_ratio": 1.544871794871795 }, { "start": 403.56, "end": 407.96, "text": "there's a security-focused variant called 3.5 Flash Cyber Built", "chunk": 0, "language": "en", "avg_logprob": -0.1060038169225057, "no_speech_prob": 1.0315117900464021e-12, "compression_ratio": 1.544871794871795 }, { "start": 407.96, "end": 410.84, "text": "to coordinate with vulnerability scanning tools,", "chunk": 0, "language": "en", "avg_logprob": -0.1060038169225057, "no_speech_prob": 1.0315117900464021e-12, "compression_ratio": 1.544871794871795 }, { "start": 410.84, "end": 414.12, "text": "and Lyria 3.5, Google's music model,", "chunk": 0, "language": "en", "avg_logprob": -0.1060038169225057, "no_speech_prob": 1.0315117900464021e-12, "compression_ratio": 1.544871794871795 }, { "start": 414.12, "end": 417.96, "text": "which can now generate tracks up to three minutes long from a text prompt.", "chunk": 0, "language": "en", "avg_logprob": -0.1060038169225057, "no_speech_prob": 1.0315117900464021e-12, "compression_ratio": 1.544871794871795 }, { "start": 417.96, "end": 420.28, "text": "Here's the honest limitation worth naming.", "chunk": 0, "language": "en", "avg_logprob": -0.1060038169225057, "no_speech_prob": 1.0315117900464021e-12, "compression_ratio": 1.544871794871795 }, { "start": 420.36, "end": 425.08, "text": "Google ships a lot of these models fast, and the naming gets confusing on purpose or not.", "chunk": 0, "language": "en", "avg_logprob": -0.1183179792810659, "no_speech_prob": 1.1114064312964245e-12, "compression_ratio": 1.578397212543554 }, { "start": 425.4, "end": 430.68, "text": "3.5, 3.6, 3.1 Pro, Flashlight, FlashCyber.", "chunk": 0, "language": "en", "avg_logprob": -0.1183179792810659, "no_speech_prob": 1.1114064312964245e-12, "compression_ratio": 1.578397212543554 }, { "start": 431.08, "end": 436.52, "text": "If you're not building on top of the API professionally, you genuinely don't need to memorize this list.", "chunk": 0, "language": "en", "avg_logprob": -0.1183179792810659, "no_speech_prob": 1.1114064312964245e-12, "compression_ratio": 1.578397212543554 }, { "start": 436.52, "end": 441.54, "text": "You just need to know the shape of it. Fast and cheap, deep reasoning, and multimodal.", "chunk": 0, "language": "en", "avg_logprob": -0.1183179792810659, "no_speech_prob": 1.1114064312964245e-12, "compression_ratio": 1.578397212543554 }, { "start": 441.84, "end": 444.88, "text": "That's really three categories wearing a lot of different name tags.", "chunk": 0, "language": "en", "avg_logprob": -0.1183179792810659, "no_speech_prob": 1.1114064312964245e-12, "compression_ratio": 1.578397212543554 }, { "start": 445.44, "end": 448.8, "text": "The modes you actually interact with models are the engine.", "chunk": 0, "language": "en", "avg_logprob": -0.1183179792810659, "no_speech_prob": 1.1114064312964245e-12, "compression_ratio": 1.578397212543554 }, { "start": 448.8, "end": 453.56, "text": "Modes are the steering wheel. Here's where things get useful for anyone who isn't a developer.", "chunk": 0, "language": "en", "avg_logprob": -0.12516483238765172, "no_speech_prob": 1.9657936303069112e-12, "compression_ratio": 1.5503355704697988 }, { "start": 454.0, "end": 459.86, "text": "AI mode inside Google search turns your search bar into a conversation. As of IO 2026,", "chunk": 0, "language": "en", "avg_logprob": -0.12516483238765172, "no_speech_prob": 1.9657936303069112e-12, "compression_ratio": 1.5503355704697988 }, { "start": 460.42, "end": 466.02, "text": "it's globally powered by Gemini 3.5 Flash, and instead of 10 blue links, you get a written", "chunk": 0, "language": "en", "avg_logprob": -0.12516483238765172, "no_speech_prob": 1.9657936303069112e-12, "compression_ratio": 1.5503355704697988 }, { "start": 466.02, "end": 470.9, "text": "answer with follow-up questions and sometimes an interactive widget built on the fly. Anyone", "chunk": 0, "language": "en", "avg_logprob": -0.12516483238765172, "no_speech_prob": 1.9657936303069112e-12, "compression_ratio": 1.5503355704697988 }, { "start": 470.9, "end": 476.0, "text": "with search can use it. No subscription required. Ask something like, what's a quick dinner with", "chunk": 0, "language": "en", "avg_logprob": -0.12516483238765172, "no_speech_prob": 1.9657936303069112e-12, "compression_ratio": 1.5503355704697988 }, { "start": 476.0, "end": 480.0, "text": "what's in my fridge and it answers in full sentences not a list of recipes", "chunk": 0, "language": "en", "avg_logprob": -0.17068599399767423, "no_speech_prob": 1.7544402345845422e-12, "compression_ratio": 1.0571428571428572 }, { "start": 480.86, "end": 483.8, "text": "DeepThink is the extra effort version of the Gemini app.", "chunk": 1, "language": "en", "avg_logprob": -0.1476876632027004, "no_speech_prob": 3.7013682049891194e-12, "compression_ratio": 1.588628762541806 }, { "start": 484.18, "end": 488.64, "text": "It spends more compute per answer to reason through harder problems step by step.", "chunk": 1, "language": "en", "avg_logprob": -0.1476876632027004, "no_speech_prob": 3.7013682049891194e-12, "compression_ratio": 1.588628762541806 }, { "start": 488.96, "end": 491.26, "text": "Google gates this one behind Google AI Ultra,", "chunk": 1, "language": "en", "avg_logprob": -0.1476876632027004, "no_speech_prob": 3.7013682049891194e-12, "compression_ratio": 1.588628762541806 }, { "start": 491.26, "end": 496.62, "text": "and it's built for genuinely difficult science or engineering questions, not everyday chat.", "chunk": 1, "language": "en", "avg_logprob": -0.12338061843599592, "no_speech_prob": 2.427435639648645e-12, "compression_ratio": 1.5611510791366907 }, { "start": 496.98, "end": 501.36, "text": "Now, deep research is where this stops being a chatbot and starts being an assistant.", "chunk": 1, "language": "en", "avg_logprob": -0.12338061843599592, "no_speech_prob": 2.427435639648645e-12, "compression_ratio": 1.5611510791366907 }, { "start": 501.8, "end": 505.92, "text": "You give it a topic, and instead of one reply, it plans a research strategy,", "chunk": 1, "language": "en", "avg_logprob": -0.12338061843599592, "no_speech_prob": 2.427435639648645e-12, "compression_ratio": 1.5611510791366907 }, { "start": 506.48, "end": 512.36, "text": "opens webpages, reads them, and, if you allow it, pulls from your own Gmail and Drive too.", "chunk": 1, "language": "en", "avg_logprob": -0.12338061843599592, "no_speech_prob": 2.427435639648645e-12, "compression_ratio": 1.5611510791366907 }, { "start": 512.68, "end": 514.12, "text": "What comes back isn't a paragraph.", "chunk": 1, "language": "en", "avg_logprob": -0.12338061843599592, "no_speech_prob": 2.427435639648645e-12, "compression_ratio": 1.5611510791366907 }, { "start": 514.66, "end": 517.74, "text": "It's a full multi-page report inside Gemini's Canvas.", "chunk": 1, "language": "en", "avg_logprob": -0.12338061843599592, "no_speech_prob": 2.427435639648645e-12, "compression_ratio": 1.5611510791366907 }, { "start": 517.74, "end": 524.76, "text": "This is the part of Gemini that actually earns the word agent, and we're coming back to why that matters in a few minutes.", "chunk": 1, "language": "en", "avg_logprob": -0.12025137166030535, "no_speech_prob": 1.2644323522242873e-12, "compression_ratio": 1.658703071672355 }, { "start": 525.18, "end": 527.4, "text": "Gemini Live is the voice and camera mode.", "chunk": 1, "language": "en", "avg_logprob": -0.12025137166030535, "no_speech_prob": 1.2644323522242873e-12, "compression_ratio": 1.658703071672355 }, { "start": 527.8, "end": 534.94, "text": "Say, hey Google, let's chat, and you're talking to it hands-free, with the option to point your camera at something and ask what it's looking at.", "chunk": 1, "language": "en", "avg_logprob": -0.12025137166030535, "no_speech_prob": 1.2644323522242873e-12, "compression_ratio": 1.658703071672355 }, { "start": 535.24, "end": 535.6, "text": "Live.", "chunk": 1, "language": "en", "avg_logprob": -0.12025137166030535, "no_speech_prob": 1.2644323522242873e-12, "compression_ratio": 1.658703071672355 }, { "start": 535.92, "end": 537.6, "text": "And Canvas is the workspace mode.", "chunk": 1, "language": "en", "avg_logprob": -0.12025137166030535, "no_speech_prob": 1.2644323522242873e-12, "compression_ratio": 1.658703071672355 }, { "start": 538.0, "end": 544.22, "text": "Type, create a quiz app about planets, and it writes the code, the interface, and the content in one pass.", "chunk": 1, "language": "en", "avg_logprob": -0.12025137166030535, "no_speech_prob": 1.2644323522242873e-12, "compression_ratio": 1.658703071672355 }, { "start": 544.7, "end": 545.78, "text": "Right there for you to edit.", "chunk": 1, "language": "en", "avg_logprob": -0.12025137166030535, "no_speech_prob": 1.2644323522242873e-12, "compression_ratio": 1.658703071672355 }, { "start": 545.78, "end": 549.12, "text": "One more worth a mention, briefly, because it's still early.", "chunk": 1, "language": "en", "avg_logprob": -0.1274623726353501, "no_speech_prob": 8.037697027049595e-13, "compression_ratio": 1.5517241379310345 }, { "start": 549.5, "end": 553.6, "text": "Gemini Spark, a personal agent announced at IO 2026,", "chunk": 1, "language": "en", "avg_logprob": -0.1274623726353501, "no_speech_prob": 8.037697027049595e-13, "compression_ratio": 1.5517241379310345 }, { "start": 554.18, "end": 557.5, "text": "meant to run continuously in the background handling things like scheduling.", "chunk": 1, "language": "en", "avg_logprob": -0.1274623726353501, "no_speech_prob": 8.037697027049595e-13, "compression_ratio": 1.5517241379310345 }, { "start": 557.98, "end": 560.14, "text": "Right now it's limited to early ultra testers.", "chunk": 1, "language": "en", "avg_logprob": -0.1274623726353501, "no_speech_prob": 8.037697027049595e-13, "compression_ratio": 1.5517241379310345 }, { "start": 560.58, "end": 562.74, "text": "So treat this one as coming, not here.", "chunk": 1, "language": "en", "avg_logprob": -0.1274623726353501, "no_speech_prob": 8.037697027049595e-13, "compression_ratio": 1.5517241379310345 }, { "start": 563.1, "end": 564.7, "text": "Quick gut check before we move on.", "chunk": 1, "language": "en", "avg_logprob": -0.1274623726353501, "no_speech_prob": 8.037697027049595e-13, "compression_ratio": 1.5517241379310345 }, { "start": 565.04, "end": 568.06, "text": "If all of that sounds like a lot of separate tools, that's fair.", "chunk": 1, "language": "en", "avg_logprob": -0.1274623726353501, "no_speech_prob": 8.037697027049595e-13, "compression_ratio": 1.5517241379310345 }, { "start": 568.3, "end": 569.26, "text": "But notice the pattern.", "chunk": 1, "language": "en", "avg_logprob": -0.1274623726353501, "no_speech_prob": 8.037697027049595e-13, "compression_ratio": 1.5517241379310345 }, { "start": 569.68, "end": 575.54, "text": "Every single one of these modes is just Gemini 3.5 or 3.6 Flash wearing a different job title.", "chunk": 1, "language": "en", "avg_logprob": -0.1274623726353501, "no_speech_prob": 8.037697027049595e-13, "compression_ratio": 1.5517241379310345 }, { "start": 575.78, "end": 578.02, "text": "You're not learning six different AIs.", "chunk": 1, "language": "en", "avg_logprob": -0.12579822540283203, "no_speech_prob": 2.4161933304817085e-12, "compression_ratio": 1.6728971962616823 }, { "start": 578.02, "end": 581.3, "text": "You're learning six different ways to ask the same brain for help.", "chunk": 1, "language": "en", "avg_logprob": -0.12579822540283203, "no_speech_prob": 2.4161933304817085e-12, "compression_ratio": 1.6728971962616823 }, { "start": 581.3, "end": 582.98, "text": "Multimodal and practice.", "chunk": 1, "language": "en", "avg_logprob": -0.12579822540283203, "no_speech_prob": 2.4161933304817085e-12, "compression_ratio": 1.6728971962616823 }, { "start": 582.98, "end": 586.1, "text": "Let's talk about what multimodal actually means day to day,", "chunk": 1, "language": "en", "avg_logprob": -0.12579822540283203, "no_speech_prob": 2.4161933304817085e-12, "compression_ratio": 1.6728971962616823 }, { "start": 586.1, "end": 588.34, "text": "because it's more than a buzzword on a slide.", "chunk": 1, "language": "en", "avg_logprob": -0.12579822540283203, "no_speech_prob": 2.4161933304817085e-12, "compression_ratio": 1.6728971962616823 }, { "start": 588.34, "end": 590.66, "text": "Gemini reads and writes text and code.", "chunk": 1, "language": "en", "avg_logprob": -0.12579822540283203, "no_speech_prob": 2.4161933304817085e-12, "compression_ratio": 1.6728971962616823 }, { "start": 590.66, "end": 592.58, "text": "Obviously, that's the baseline.", "chunk": 1, "language": "en", "avg_logprob": -0.12579822540283203, "no_speech_prob": 2.4161933304817085e-12, "compression_ratio": 1.6728971962616823 }, { "start": 592.58, "end": 596.02, "text": "But drop a photo into a chat and ask it to caption or edit it,", "chunk": 1, "language": "en", "avg_logprob": -0.12579822540283203, "no_speech_prob": 2.4161933304817085e-12, "compression_ratio": 1.6728971962616823 }, { "start": 596.02, "end": 597.7, "text": "and Nano Banana handles that.", "chunk": 1, "language": "en", "avg_logprob": -0.12579822540283203, "no_speech_prob": 2.4161933304817085e-12, "compression_ratio": 1.6728971962616823 }, { "start": 597.7, "end": 599.7, "text": "Ask it to speak and answer out loud,", "chunk": 1, "language": "en", "avg_logprob": -0.12579822540283203, "no_speech_prob": 2.4161933304817085e-12, "compression_ratio": 1.6728971962616823 }, { "start": 599.7, "end": 603.06, "text": "and Gemini's audio models generate that voice on the spot", "chunk": 1, "language": "en", "avg_logprob": -0.12579822540283203, "no_speech_prob": 2.4161933304817085e-12, "compression_ratio": 1.6728971962616823 }, { "start": 603.06, "end": 605.7, "text": "with actual control over tone and pacing.", "chunk": 1, "language": "en", "avg_logprob": -0.12579822540283203, "no_speech_prob": 2.4161933304817085e-12, "compression_ratio": 1.6728971962616823 }, { "start": 605.78, "end": 608.78, "text": "Ask for a short video, and VO builds one from scratch.", "chunk": 1, "language": "en", "avg_logprob": -0.12893340505402665, "no_speech_prob": 1.579728860744145e-12, "compression_ratio": 1.5528169014084507 }, { "start": 609.16, "end": 610.84, "text": "Ask it to edit an existing clip.", "chunk": 1, "language": "en", "avg_logprob": -0.12893340505402665, "no_speech_prob": 1.579728860744145e-12, "compression_ratio": 1.5528169014084507 }, { "start": 611.28, "end": 615.74, "text": "Swap the sky, change the style, and that's a separate tool called Gemini Omni,", "chunk": 1, "language": "en", "avg_logprob": -0.12893340505402665, "no_speech_prob": 1.579728860744145e-12, "compression_ratio": 1.5528169014084507 }, { "start": 616.04, "end": 618.28, "text": "doing frame-by-frame editing by voice command.", "chunk": 1, "language": "en", "avg_logprob": -0.12893340505402665, "no_speech_prob": 1.579728860744145e-12, "compression_ratio": 1.5528169014084507 }, { "start": 618.7, "end": 624.22, "text": "Inside Google Docs and Sheets, the same underlying models can draft a document from your meeting notes", "chunk": 1, "language": "en", "avg_logprob": -0.12893340505402665, "no_speech_prob": 1.579728860744145e-12, "compression_ratio": 1.5528169014084507 }, { "start": 624.22, "end": 629.28, "text": "or build a spreadsheet out of a pile of invoices, complete with formulas and charts,", "chunk": 1, "language": "en", "avg_logprob": -0.12893340505402665, "no_speech_prob": 1.579728860744145e-12, "compression_ratio": 1.5528169014084507 }, { "start": 629.5, "end": 631.76, "text": "not just raw numbers dumped into cells.", "chunk": 1, "language": "en", "avg_logprob": -0.12893340505402665, "no_speech_prob": 1.579728860744145e-12, "compression_ratio": 1.5528169014084507 }, { "start": 631.76, "end": 638.52, "text": "In Slides, hand it a list of bullet points and it can lay out an actual deck, not just text on blank slides.", "chunk": 1, "language": "en", "avg_logprob": -0.14819806151919895, "no_speech_prob": 8.132088210287269e-13, "compression_ratio": 1.6317460317460317 }, { "start": 638.94, "end": 643.9, "text": "And through Gemini Live's camera mode, you can point your phone at a menu in a language you don't speak", "chunk": 1, "language": "en", "avg_logprob": -0.14819806151919895, "no_speech_prob": 8.132088210287269e-13, "compression_ratio": 1.6317460317460317 }, { "start": 643.9, "end": 646.94, "text": "and get a live translation overlaid on what you're looking at,", "chunk": 1, "language": "en", "avg_logprob": -0.14819806151919895, "no_speech_prob": 8.132088210287269e-13, "compression_ratio": 1.6317460317460317 }, { "start": 647.14, "end": 650.26, "text": "or ask it to identify an object it's looking at through the lens.", "chunk": 1, "language": "en", "avg_logprob": -0.14819806151919895, "no_speech_prob": 8.132088210287269e-13, "compression_ratio": 1.6317460317460317 }, { "start": 650.62, "end": 651.6, "text": "No typing involved.", "chunk": 1, "language": "en", "avg_logprob": -0.14819806151919895, "no_speech_prob": 8.132088210287269e-13, "compression_ratio": 1.6317460317460317 }, { "start": 651.92, "end": 653.2, "text": "Here's the part worth remembering.", "chunk": 1, "language": "en", "avg_logprob": -0.14819806151919895, "no_speech_prob": 8.132088210287269e-13, "compression_ratio": 1.6317460317460317 }, { "start": 653.64, "end": 654.72, "text": "You never pick the model.", "chunk": 1, "language": "en", "avg_logprob": -0.14819806151919895, "no_speech_prob": 8.132088210287269e-13, "compression_ratio": 1.6317460317460317 }, { "start": 655.08, "end": 656.18, "text": "You just say what you want.", "chunk": 1, "language": "en", "avg_logprob": -0.14819806151919895, "no_speech_prob": 8.132088210287269e-13, "compression_ratio": 1.6317460317460317 }, { "start": 656.48, "end": 657.62, "text": "Make this an infographic.", "chunk": 1, "language": "en", "avg_logprob": -0.14819806151919895, "no_speech_prob": 8.132088210287269e-13, "compression_ratio": 1.6317460317460317 }, { "start": 658.14, "end": 658.94, "text": "Translate this.", "chunk": 1, "language": "en", "avg_logprob": -0.14819806151919895, "no_speech_prob": 8.132088210287269e-13, "compression_ratio": 1.6317460317460317 }, { "start": 659.36, "end": 660.24, "text": "Write this in Python.", "chunk": 1, "language": "en", "avg_logprob": -0.14819806151919895, "no_speech_prob": 8.132088210287269e-13, "compression_ratio": 1.6317460317460317 }, { "start": 660.24, "end": 664.94, "text": "and Gemini quietly roots the request to whichever model actually does that job.", "chunk": 1, "language": "en", "avg_logprob": -0.16228645266467379, "no_speech_prob": 1.3565261110584648e-12, "compression_ratio": 1.6079734219269104 }, { "start": 665.28, "end": 667.3, "text": "That's the design philosophy in one sentence.", "chunk": 1, "language": "en", "avg_logprob": -0.16228645266467379, "no_speech_prob": 1.3565261110584648e-12, "compression_ratio": 1.6079734219269104 }, { "start": 667.88, "end": 671.16, "text": "One platform, and it decides the plumbing so you don't have to.", "chunk": 1, "language": "en", "avg_logprob": -0.16228645266467379, "no_speech_prob": 1.3565261110584648e-12, "compression_ratio": 1.6079734219269104 }, { "start": 671.5, "end": 672.92, "text": "Where Gemini actually lives.", "chunk": 1, "language": "en", "avg_logprob": -0.16228645266467379, "no_speech_prob": 1.3565261110584648e-12, "compression_ratio": 1.6079734219269104 }, { "start": 673.28, "end": 675.04, "text": "This is the part that's easy to underestimate.", "chunk": 1, "language": "en", "avg_logprob": -0.16228645266467379, "no_speech_prob": 1.3565261110584648e-12, "compression_ratio": 1.6079734219269104 }, { "start": 675.68, "end": 677.56, "text": "Gemini isn't confined to one app.", "chunk": 1, "language": "en", "avg_logprob": -0.16228645266467379, "no_speech_prob": 1.3565261110584648e-12, "compression_ratio": 1.6079734219269104 }, { "start": 678.0, "end": 680.24, "text": "It's spread across nearly everything Google ships.", "chunk": 1, "language": "en", "avg_logprob": -0.16228645266467379, "no_speech_prob": 1.3565261110584648e-12, "compression_ratio": 1.6079734219269104 }, { "start": 680.7, "end": 682.98, "text": "In search, it's AI mode, already covered.", "chunk": 1, "language": "en", "avg_logprob": -0.16228645266467379, "no_speech_prob": 1.3565261110584648e-12, "compression_ratio": 1.6079734219269104 }, { "start": 683.46, "end": 686.86, "text": "In Gmail, it's behind Smart Compose and Auto Reply suggestions.", "chunk": 1, "language": "en", "avg_logprob": -0.16228645266467379, "no_speech_prob": 1.3565261110584648e-12, "compression_ratio": 1.6079734219269104 }, { "start": 687.32, "end": 688.9, "text": "In Docs, Sheets and Slides.", "chunk": 1, "language": "en", "avg_logprob": -0.16228645266467379, "no_speech_prob": 1.3565261110584648e-12, "compression_ratio": 1.6079734219269104 }, { "start": 688.9, "end": 695.36, "text": "Ultra and Pro subscribers get Gemini drafting text, building formulas, and designing slide layouts,", "chunk": 1, "language": "en", "avg_logprob": -0.12503703294602114, "no_speech_prob": 2.0124171426494364e-12, "compression_ratio": 1.6032786885245902 }, { "start": 695.88, "end": 698.24, "text": "pulling context from your own files when you let it.", "chunk": 1, "language": "en", "avg_logprob": -0.12503703294602114, "no_speech_prob": 2.0124171426494364e-12, "compression_ratio": 1.6032786885245902 }, { "start": 698.5, "end": 701.52, "text": "In Drive, it can find and summarize documents for you.", "chunk": 1, "language": "en", "avg_logprob": -0.12503703294602114, "no_speech_prob": 2.0124171426494364e-12, "compression_ratio": 1.6032786885245902 }, { "start": 701.92, "end": 704.32, "text": "In Google Meet, it's doing live caption translation.", "chunk": 1, "language": "en", "avg_logprob": -0.12503703294602114, "no_speech_prob": 2.0124171426494364e-12, "compression_ratio": 1.6032786885245902 }, { "start": 704.82, "end": 709.16, "text": "On Android, especially Pixel devices, it's baked straight into the Voice Assistant,", "chunk": 1, "language": "en", "avg_logprob": -0.12503703294602114, "no_speech_prob": 2.0124171426494364e-12, "compression_ratio": 1.6032786885245902 }, { "start": 709.54, "end": 713.66, "text": "and there's a Chrome extension that lets the browser send page content straight to Gemini", "chunk": 1, "language": "en", "avg_logprob": -0.12503703294602114, "no_speech_prob": 2.0124171426494364e-12, "compression_ratio": 1.6032786885245902 }, { "start": 713.66, "end": 716.42, "text": "so you can ask questions about whatever tab you're on.", "chunk": 1, "language": "en", "avg_logprob": -0.12503703294602114, "no_speech_prob": 2.0124171426494364e-12, "compression_ratio": 1.6032786885245902 }, { "start": 716.42, "end": 721.68, "text": "And for developers, all of it is exposed through Google AI Studio and the Gemini API,", "chunk": 1, "language": "en", "avg_logprob": -0.11233553741917465, "no_speech_prob": 1.289348510770394e-12, "compression_ratio": 1.632258064516129 }, { "start": 722.22, "end": 727.04, "text": "plus a newer platform called Anti-Gravity for building multi-agent workflows on top of it.", "chunk": 1, "language": "en", "avg_logprob": -0.11233553741917465, "no_speech_prob": 1.289348510770394e-12, "compression_ratio": 1.632258064516129 }, { "start": 727.28, "end": 729.08, "text": "The strategic point here isn't subtle.", "chunk": 1, "language": "en", "avg_logprob": -0.11233553741917465, "no_speech_prob": 1.289348510770394e-12, "compression_ratio": 1.632258064516129 }, { "start": 729.46, "end": 732.32, "text": "Google isn't trying to win the best standalone chatbot argument.", "chunk": 1, "language": "en", "avg_logprob": -0.11233553741917465, "no_speech_prob": 1.289348510770394e-12, "compression_ratio": 1.632258064516129 }, { "start": 732.7, "end": 735.94, "text": "It's trying to make sure you're never more than one product away from Gemini,", "chunk": 1, "language": "en", "avg_logprob": -0.11233553741917465, "no_speech_prob": 1.289348510770394e-12, "compression_ratio": 1.632258064516129 }, { "start": 736.3, "end": 739.38, "text": "no matter what you're doing on a Google device or in a Google app.", "chunk": 1, "language": "en", "avg_logprob": -0.11233553741917465, "no_speech_prob": 1.289348510770394e-12, "compression_ratio": 1.632258064516129 }, { "start": 739.78, "end": 742.0, "text": "Agents, the part that actually matters.", "chunk": 1, "language": "en", "avg_logprob": -0.11233553741917465, "no_speech_prob": 1.289348510770394e-12, "compression_ratio": 1.632258064516129 }, { "start": 742.38, "end": 744.22, "text": "Now here's the shift I promised earlier.", "chunk": 1, "language": "en", "avg_logprob": -0.11233553741917465, "no_speech_prob": 1.289348510770394e-12, "compression_ratio": 1.632258064516129 }, { "start": 744.22, "end": 747.24, "text": "the one that actually changes what this platform is for.", "chunk": 1, "language": "en", "avg_logprob": -0.15245236090894015, "no_speech_prob": 9.962731594564955e-13, "compression_ratio": 1.5703971119133573 }, { "start": 747.64, "end": 751.16, "text": "Everything so far has been ask a question, get an answer.", "chunk": 1, "language": "en", "avg_logprob": -0.15245236090894015, "no_speech_prob": 9.962731594564955e-13, "compression_ratio": 1.5703971119133573 }, { "start": 751.56, "end": 754.48, "text": "Agents are Google trying to move Gemini past that entirely.", "chunk": 1, "language": "en", "avg_logprob": -0.15245236090894015, "no_speech_prob": 9.962731594564955e-13, "compression_ratio": 1.5703971119133573 }, { "start": 754.9, "end": 760.62, "text": "Deep research is the clearest example already live, plan, browse, synthesize, write,", "chunk": 1, "language": "en", "avg_logprob": -0.15245236090894015, "no_speech_prob": 9.962731594564955e-13, "compression_ratio": 1.5703971119133573 }, { "start": 760.96, "end": 762.58, "text": "without you babysitting every step.", "chunk": 1, "language": "en", "avg_logprob": -0.15245236090894015, "no_speech_prob": 9.962731594564955e-13, "compression_ratio": 1.5703971119133573 }, { "start": 762.98, "end": 766.92, "text": "Spark is the early, still limited attempt at a persistent personal agent", "chunk": 1, "language": "en", "avg_logprob": -0.15245236090894015, "no_speech_prob": 9.962731594564955e-13, "compression_ratio": 1.5703971119133573 }, { "start": 766.92, "end": 768.82, "text": "running continuously in the background.", "chunk": 1, "language": "en", "avg_logprob": -0.15245236090894015, "no_speech_prob": 9.962731594564955e-13, "compression_ratio": 1.5703971119133573 }, { "start": 769.28, "end": 770.44, "text": "And on the developer side,", "chunk": 1, "language": "en", "avg_logprob": -0.15245236090894015, "no_speech_prob": 9.962731594564955e-13, "compression_ratio": 1.5703971119133573 }, { "start": 770.44, "end": 775.92, "text": "anti-gravity lets companies build coordinated teams of sub-agents. Google's own blog post gave", "chunk": 1, "language": "en", "avg_logprob": -0.05714532785248338, "no_speech_prob": 1.068923270210187e-12, "compression_ratio": 1.5742574257425743 }, { "start": 775.92, "end": 781.32, "text": "an example of businesses running parallel agents to analyze data at scale, rather than one model", "chunk": 1, "language": "en", "avg_logprob": -0.05714532785248338, "no_speech_prob": 1.068923270210187e-12, "compression_ratio": 1.5742574257425743 }, { "start": 781.32, "end": 786.28, "text": "doing everything sequentially. Picture the difference in practice. The old way. You ask", "chunk": 1, "language": "en", "avg_logprob": -0.05714532785248338, "no_speech_prob": 1.068923270210187e-12, "compression_ratio": 1.5742574257425743 }, { "start": 786.28, "end": 791.8, "text": "Gemini, what should I know before a trip to Japan? And it gives you a paragraph. The agent way. You", "chunk": 1, "language": "en", "avg_logprob": -0.05714532785248338, "no_speech_prob": 1.068923270210187e-12, "compression_ratio": 1.5742574257425743 }, { "start": 791.8, "end": 797.94, "text": "say, plan my trip to Japan. And it checks flights, compares hotel options, and comes back with an", "chunk": 1, "language": "en", "avg_logprob": -0.05714532785248338, "no_speech_prob": 1.068923270210187e-12, "compression_ratio": 1.5742574257425743 }, { "start": 797.94, "end": 803.04, "text": "actual itinerary, pausing to confirm with you before it books anything. That's the same underlying", "chunk": 1, "language": "en", "avg_logprob": -0.06607280671596527, "no_speech_prob": 2.343493888529169e-12, "compression_ratio": 1.6474164133738602 }, { "start": 803.04, "end": 807.66, "text": "model, just given permission to take more than one step before handing control back to you.", "chunk": 1, "language": "en", "avg_logprob": -0.06607280671596527, "no_speech_prob": 2.343493888529169e-12, "compression_ratio": 1.6474164133738602 }, { "start": 808.0, "end": 811.62, "text": "None of this is science fiction anymore, and none of it is fully finished either.", "chunk": 1, "language": "en", "avg_logprob": -0.06607280671596527, "no_speech_prob": 2.343493888529169e-12, "compression_ratio": 1.6474164133738602 }, { "start": 811.94, "end": 817.14, "text": "That's the honest read. Deep research genuinely works today. Spark is still in early testing,", "chunk": 1, "language": "en", "avg_logprob": -0.06607280671596527, "no_speech_prob": 2.343493888529169e-12, "compression_ratio": 1.6474164133738602 }, { "start": 817.14, "end": 822.08, "text": "but the direction is unmistakable. Google wants Gemini to eventually take a task,", "chunk": 1, "language": "en", "avg_logprob": -0.06607280671596527, "no_speech_prob": 2.343493888529169e-12, "compression_ratio": 1.6474164133738602 }, { "start": 822.5, "end": 827.4, "text": "break it into steps, and execute most of them without you typing a follow-up for every single", "chunk": 1, "language": "en", "avg_logprob": -0.06607280671596527, "no_speech_prob": 2.343493888529169e-12, "compression_ratio": 1.6474164133738602 }, { "start": 827.4, "end": 832.64, "text": "one. What actually makes Gemini different? So how does this stack up against everyone else building", "chunk": 1, "language": "en", "avg_logprob": -0.08343253176436465, "no_speech_prob": 2.29789386303747e-12, "compression_ratio": 1.5594855305466238 }, { "start": 832.64, "end": 838.06, "text": "the same kind of thing? Let's be balanced here, because Google's advantages are real, but so are", "chunk": 1, "language": "en", "avg_logprob": -0.08343253176436465, "no_speech_prob": 2.29789386303747e-12, "compression_ratio": 1.5594855305466238 }, { "start": 838.06, "end": 844.2, "text": "its weak spots. The clearest edge is data. Gemini can pull from live search results, maps, Gmail,", "chunk": 1, "language": "en", "avg_logprob": -0.08343253176436465, "no_speech_prob": 2.29789386303747e-12, "compression_ratio": 1.5594855305466238 }, { "start": 844.56, "end": 850.58, "text": "and Drive in ways that a closed, sandbox chatbot simply can't match without plugins bolted on.", "chunk": 1, "language": "en", "avg_logprob": -0.08343253176436465, "no_speech_prob": 2.29789386303747e-12, "compression_ratio": 1.5594855305466238 }, { "start": 850.58, "end": 856.38, "text": "The second edge is reach. Every Android phone is a potential Gemini client, and every workspace", "chunk": 1, "language": "en", "avg_logprob": -0.08343253176436465, "no_speech_prob": 2.29789386303747e-12, "compression_ratio": 1.5594855305466238 }, { "start": 856.38, "end": 861.62, "text": "business account already has it available. No competitor has that kind of built-in distribution.", "chunk": 1, "language": "en", "avg_logprob": -0.0880875954261193, "no_speech_prob": 3.816163097331016e-12, "compression_ratio": 1.5439189189189189 }, { "start": 862.1, "end": 868.12, "text": "And on raw benchmarks, Gemini 3 Pro topped the LM Arena leaderboard, which, regardless of how much", "chunk": 1, "language": "en", "avg_logprob": -0.0880875954261193, "no_speech_prob": 3.816163097331016e-12, "compression_ratio": 1.5439189189189189 }, { "start": 868.12, "end": 872.68, "text": "weight you put on any single leaderboard, says Google's infrastructure and DeepMind's research", "chunk": 1, "language": "en", "avg_logprob": -0.0880875954261193, "no_speech_prob": 3.816163097331016e-12, "compression_ratio": 1.5439189189189189 }, { "start": 872.68, "end": 878.52, "text": "are producing real, top-tier results, not just hype. But, and this matters for credibility,", "chunk": 1, "language": "en", "avg_logprob": -0.0880875954261193, "no_speech_prob": 3.816163097331016e-12, "compression_ratio": 1.5439189189189189 }, { "start": 879.06, "end": 882.42, "text": "Google is genuinely more conservative about rollout than some competitors.", "chunk": 1, "language": "en", "avg_logprob": -0.0880875954261193, "no_speech_prob": 3.816163097331016e-12, "compression_ratio": 1.5439189189189189 }, { "start": 882.42, "end": 887.42, "text": "DeepThink and Spark are still gated behind ultra subscriptions or limited testing, while", "chunk": 1, "language": "en", "avg_logprob": -0.14942777045419284, "no_speech_prob": 1.7412900548546828e-12, "compression_ratio": 1.5317725752508362 }, { "start": 887.42, "end": 890.58, "text": "some rivals ship new capabilities to everyone at once.", "chunk": 1, "language": "en", "avg_logprob": -0.14942777045419284, "no_speech_prob": 1.7412900548546828e-12, "compression_ratio": 1.5317725752508362 }, { "start": 890.58, "end": 897.66, "text": "And the tier structure itself — free, pro, ultra, API pricing — can be genuinely confusing", "chunk": 1, "language": "en", "avg_logprob": -0.14942777045419284, "no_speech_prob": 1.7412900548546828e-12, "compression_ratio": 1.5317725752508362 }, { "start": 897.66, "end": 900.64, "text": "next to a simpler flat subscription from a competitor.", "chunk": 1, "language": "en", "avg_logprob": -0.14942777045419284, "no_speech_prob": 1.7412900548546828e-12, "compression_ratio": 1.5317725752508362 }, { "start": 900.64, "end": 904.2, "text": "If you've ever opened the Gemini pricing page and closed it five minutes later still", "chunk": 1, "language": "en", "avg_logprob": -0.14942777045419284, "no_speech_prob": 1.7412900548546828e-12, "compression_ratio": 1.5317725752508362 }, { "start": 904.2, "end": 907.26, "text": "unsure which plan you need, that's not just you.", "chunk": 1, "language": "en", "avg_logprob": -0.14942777045419284, "no_speech_prob": 1.7412900548546828e-12, "compression_ratio": 1.5317725752508362 }, { "start": 907.26, "end": 908.62, "text": "Where this is actually headed.", "chunk": 1, "language": "en", "avg_logprob": -0.14942777045419284, "no_speech_prob": 1.7412900548546828e-12, "compression_ratio": 1.5317725752508362 }, { "start": 908.62, "end": 912.6, "text": "A few things are confirmed, and a few are still rumor, and it's worth keeping those separate.", "chunk": 1, "language": "en", "avg_logprob": -0.10849683625357491, "no_speech_prob": 1.4212353082604379e-12, "compression_ratio": 1.5780730897009967 }, { "start": 912.96, "end": 918.36, "text": "Confirmed. Gemini 3.5 Pro is currently in partner testing with a public release expected soon,", "chunk": 1, "language": "en", "avg_logprob": -0.10849683625357491, "no_speech_prob": 1.4212353082604379e-12, "compression_ratio": 1.5780730897009967 }, { "start": 918.36, "end": 923.86, "text": "and Google has already started training on Gemini 4, according to its own July 2026 announcement,", "chunk": 1, "language": "en", "avg_logprob": -0.10849683625357491, "no_speech_prob": 1.4212353082604379e-12, "compression_ratio": 1.5780730897009967 }, { "start": 924.2, "end": 928.56, "text": "though there's no public timeline for that yet. Workspace AI rollout continues expanding,", "chunk": 1, "language": "en", "avg_logprob": -0.10849683625357491, "no_speech_prob": 1.4212353082604379e-12, "compression_ratio": 1.5780730897009967 }, { "start": 929.08, "end": 934.3, "text": "and Gemini Live's regional language support keeps growing. Speculative, and worth labeling clearly", "chunk": 1, "language": "en", "avg_logprob": -0.10849683625357491, "no_speech_prob": 1.4212353082604379e-12, "compression_ratio": 1.5780730897009967 }, { "start": 934.3, "end": 940.0, "text": "as such. There's talk of a dedicated on-device AI chip for future Pixel phones, and some experimental", "chunk": 1, "language": "en", "avg_logprob": -0.08736178792756179, "no_speech_prob": 3.503608644525591e-12, "compression_ratio": 1.5224358974358974 }, { "start": 940.0, "end": 945.1, "text": "deep-mind research around 3D avatars and world simulation that hasn't shipped as a product.", "chunk": 1, "language": "en", "avg_logprob": -0.08736178792756179, "no_speech_prob": 3.503608644525591e-12, "compression_ratio": 1.5224358974358974 }, { "start": 945.48, "end": 950.5, "text": "Treat both of those as possible, not coming. Nothing official has confirmed either one.", "chunk": 1, "language": "en", "avg_logprob": -0.08736178792756179, "no_speech_prob": 3.503608644525591e-12, "compression_ratio": 1.5224358974358974 }, { "start": 950.82, "end": 956.1, "text": "The verdict. So where does that leave things? Gemini in 2026 isn't a chatbot you occasionally", "chunk": 1, "language": "en", "avg_logprob": -0.08736178792756179, "no_speech_prob": 3.503608644525591e-12, "compression_ratio": 1.5224358974358974 }, { "start": 956.1, "end": 962.56, "text": "open. It's an AI layer Google has threaded through search, Gmail, your documents, and increasingly,", "chunk": 1, "language": "en", "avg_logprob": -0.08736178792756179, "no_speech_prob": 3.503608644525591e-12, "compression_ratio": 1.5224358974358974 }, { "start": 962.56, "end": 967.38, "text": "your phone itself. The models handle the thinking, the modes handle how you ask,", "chunk": 1, "language": "en", "avg_logprob": -0.08330550988515219, "no_speech_prob": 5.220021463692159e-12, "compression_ratio": 1.7651006711409396 }, { "start": 967.72, "end": 971.64, "text": "and agents like Deep Research are the clearest sign of where all of it is actually heading.", "chunk": 1, "language": "en", "avg_logprob": -0.08330550988515219, "no_speech_prob": 5.220021463692159e-12, "compression_ratio": 1.7651006711409396 }, { "start": 972.02, "end": 976.08, "text": "If there's one thing worth trying this week, it's Deep Research on something you'd normally", "chunk": 1, "language": "en", "avg_logprob": -0.08330550988515219, "no_speech_prob": 5.220021463692159e-12, "compression_ratio": 1.7651006711409396 }, { "start": 976.08, "end": 980.1, "text": "spend an evening looking into yourself, and actually watching it work instead of just", "chunk": 1, "language": "en", "avg_logprob": -0.08330550988515219, "no_speech_prob": 5.220021463692159e-12, "compression_ratio": 1.7651006711409396 }, { "start": 980.1, "end": 983.94, "text": "reading the final report. Drop a comment with which piece of this surprised you most,", "chunk": 1, "language": "en", "avg_logprob": -0.08330550988515219, "no_speech_prob": 5.220021463692159e-12, "compression_ratio": 1.7651006711409396 }, { "start": 984.3, "end": 988.76, "text": "the model lineup, the agent side, or just how much of this you were already using without", "chunk": 1, "language": "en", "avg_logprob": -0.08330550988515219, "no_speech_prob": 5.220021463692159e-12, "compression_ratio": 1.7651006711409396 }, { "start": 988.76, "end": 993.34, "text": "realizing it. I'll be back soon with a deeper breakdown on how deep research actually performs", "chunk": 1, "language": "en", "avg_logprob": -0.07252537926962209, "no_speech_prob": 6.927948882975987e-13, "compression_ratio": 1.3458646616541354 }, { "start": 993.34, "end": 997.12, "text": "against a real research task. Thanks for watching, and I'll see you in the next one.", "chunk": 1, "language": "en", "avg_logprob": -0.07252537926962209, "no_speech_prob": 6.927948882975987e-13, "compression_ratio": 1.3458646616541354 } ], "segmented_transcription": { "status": "SEGMENTED_TRANSCRIPT_OK", "created_at": "2026-08-17T16:43:44", "manifest": "/Users/sagawa/AI_WORK/video_notes/url/20260817_164301__Google_Gemini_Explained_80_of_Google_Gemini_You_Never_Use_2026_Guide/segmented_work/chunk_manifest.json", "audio": "/Users/sagawa/AI_WORK/video_notes/url/20260817_164301__Google_Gemini_Explained_80_of_Google_Gemini_You_Never_Use_2026_Guide/audio_16k_mono.wav", "audio_sha256": "a2271f68c5c7e86e6b79f1093661debbb480adca69083224e126be96abe14a5d", "audio_duration": 997.248, "model": "mlx-community/whisper-large-v3-turbo", "language_hint": "en", "condition_on_previous_text": false, "chunk_count": 2, "resumed_chunks": 0, "merged_segment_count": 231, "final_segment_end": 997.12, "global_repeat_run": 1, "global_repeat_text": "You've probably typed something into Gemini, gotten an answer, and closed the tab.", "tail_repeat_run": 1, "tail_repeat_text": "business account already has it available. No competitor has that kind of built-in distribution.", "hallucination_region_count": 0, "hallucination_regions": [], "qc_ok": true, "qc_errors": [] } }