00:00:00There are way too many AI tools in 2026, and it has become impossible to figure out
00:00:05which ones are actually worth your time and money. And because I have nothing better to do,
00:00:10I have tested a ton of them over the last few months. And today I'm going to be ranking all
00:00:15of them. I will be covering everything from basic chat bots to full fledged coding agents and
00:00:20everything in between. So by the end of this video, you will know which ones you should actually focus
00:00:27on. So we're going to be ranking this using the standard tier list format. S tier is the best,
00:00:32D tier is the worst, and all these tools will be grouped based on their function.
00:00:36Now, when it comes to grading the specific tool, we will look at it through two lenses. One is how
00:00:42does this rank versus the other tools in its class? So how does N8N compare to other no code automation
00:00:49tools like Make or Zapier? But we will also take a look at it at the macro level because you can be
00:00:56best in class and something like no code automation. But does that mean you're automatically S tier
00:01:00because you beat everyone in your category? Not necessarily because this is all kind of tied
00:01:05together. These tools play off one another. What was great six months ago might be having issues
00:01:11today. So when I rank your favorite tools D tier, make sure to leave a comment and let me know how
00:01:14much you hate my subjective and very vague grading criteria. I'll make sure to keep that in mind next
00:01:20time. So let's kick this off with the chatbots. And the first one we are going to rank is ChatGPT. Now,
00:01:26this is where a lot of us first got our start. And by the numbers, most people still are here.
00:01:31At a macro level, ChatGPT is definitely going up in the rankings for me because their models
00:01:35are so solid. 5.4 is a great model. However, when we're talking about the chatbot itself,
00:01:40I come away just feeling meh, it's fine. It's become the ultimate middle of the road chatbot.
00:01:47You're not going to hate the outputs. You're not going to love the outputs. And unfortunately,
00:01:51if you suck at prompting, you're going to get very generic responses. We all know how to recognize
00:01:56ChatGPT writing at this point. But for 20 bucks a month, I can get a lot out of it. And it has a few
00:02:00bells and whistles like image generation. And the new image generation model is actually really good.
00:02:04And speaking of image generation, let's talk about another chatbot. And that is Gemini. Now, Gemini,
00:02:10I'll tell you right away, A tier. I think if you are just the type of person who's just going to
00:02:15spend 20 bucks a month on an AI chatbot, you just want a daily driver. Gemini is your best bet by
00:02:21far. It's better than Claude in that regard. And what do I mean by that? You're going to get enough
00:02:25usage out of it as the average person. On top of that, Gemini is going to give me image generation
00:02:31and nano banana pros best in class. I can get video generation. And importantly, what it does better
00:02:36than any other chat model is handle videos itself. I can give Gemini YouTube videos. I can give it
00:02:42short form videos, reels, TikToks, whatever, and it can actually analyze it. The other ones
00:02:46really struggle in that department or can't do it at all. Now, in a vacuum, do I think 3.1 is
00:02:50better than 5.4 or Opus like we're going to talk about? No. But I think holistically, 20 bucks a
00:02:55month, daily driver, you should be using Gemini. And in comes Anthropic. So where do we put Claude?
00:03:00This is tough because when we talk about Claude and Opus 4.7, Opus 4.7 is the best model in the
00:03:08game right now, at least in my opinion. And I think it's, Anthropic is the best out of the
00:03:12three major players. However, if we're talking purely the chat bot and you are on a $20 a month
00:03:18plan, you could argue it's C tier. Purely, purely based on the usage issues. If you're on a $20 a
00:03:26month plan and you're trying to use Opus for any real projects or have it actually generate
00:03:31anything, you're going to hit a wall so fast and you're going to ask yourself, what am I doing?
00:03:35Because for 20 bucks a month, I can probably get an output that's a 95% solution and I can
00:03:41do that a million times. Meanwhile, Anthropic throttles your usage pretty heavily if you're
00:03:45not on a $100 or $200 a month plan. Furthermore, when it comes to stuff like Sauna and Opus,
00:03:50all it can do is text and code and these artifacts. I can't do images. I can't do video. It's just text.
00:03:58So if you are the average AI user who's just using a chat bot, I would argue it's C tier.
00:04:05However, if you're on a $100 a month plan or a $200 a month plan, S tier. Unless you're trying
00:04:14to do stuff with like video or images. So when it comes to Claude, very use case dependent and very
00:04:20usage dependent. So for your average guy coming off the street, just wants to know what AI should pay
00:04:2520 bucks a month for C tier. But if you're serious about this, you have a $100 a month or $200 a month
00:04:31plan S tier. And then there was Grok. Where do we put Grok? Well, it's worse than all these at coding
00:04:37tasks. It has a much worse image model than CAD GPT or Gemini. It's video model is worse than Gemini's
00:04:44and really all it has going for it is less guardrails and it's pretty good out of the box
00:04:50in terms of figuring out what's going on in the real world at that very moment.
00:04:55So it's great if you're Twitter brained, maybe if you're Twitter brained, it's B plus tier. For
00:05:00your average user, C tier. To be totally honest with you. And when we talk about, oh, it's great at
00:05:06handling video, like I can feed it a Twitter thing, it knows what's going on. Again, if you're like
00:05:10hyper-focused on Twitter functionality, sure. Maybe A tier. Average user, C tier at best. You would
00:05:17never recommend this to someone who isn't spending four hours a day on Twitter. Now, before we jump
00:05:21into the coding agent rankings, I just want to give a quick plug for my Claude code masterclass.
00:05:26It just came out last month and it is the number one way to go from zero to AI dev, especially if
00:05:31you don't come from a technical background. I update this every week and lately we've been really
00:05:35focusing on how to build an agentic OS that is powered by Claude code and brings in stuff like
00:05:42Obsidian for memory and GWS to connect it to stuff like your Google suite. So if you're interested in
00:05:48that, you can find it inside of chase AI plus there will be a link in the pin comment. So that's it for
00:05:52the chat bots. Now let's move on to the coding agents or these coding harnesses like code X and
00:05:57Claude code. And speaking of Claude code, that will be the first one we talk about now. Claude code S
00:06:03tier for me. However, same sort of conversation as before. What are we talking about in terms of
00:06:10usage? If you think you can show up to Claude code with a $20 a month plan and get anything out of it,
00:06:15you're going to be mistaken. I'm going to keep it at S tier here because if you're using Claude code
00:06:19and you're serious about building, you're going to have a hundred dollar a month plan, bare minimum,
00:06:24or $200 a month. And we can sit here and complain about the usages and all of this,
00:06:29but the fact remains there's a reason this has become essentially an industry standard.
00:06:33It is good. And you have to remember even at a hundred dollar or $200 a month price point,
00:06:38you still get like a 90% reduction on what your normal token usage would cost if you were using
00:06:44the API's for something like Opus 4.7. Now, is there a very loud contingent of people who will
00:06:50say, Hey, Claude code has been nerfed. Opus 4.6 has been nerfed. There's problems with 4.7. Yes.
00:06:55But I think that has a lot more to do with the realities of online discussion about any tool
00:07:02that gets popular. And frankly, a number of skill issues than it is about some secret plot to nerf
00:07:08Claude code. However, just because Claude code is S tier, that doesn't mean it's the only game in
00:07:12town because if you really are of the opinion that anthropic is going downhill and these Opus 4.6 and
00:07:194.7 releases have been substandard. Well, you have options, namely codex, which is open AIs, you know,
00:07:26makers of chat GPTs version of Claude code. And I'll tell you right now it's S tier as well. And
00:07:31it has some serious advantages over Claude code. The primary one being the usage. I can get way more
00:07:38tokens out of codex than I can at a Claude code. On top of that, open AIs models are good. Whatever
00:07:45you're doing with Opus 4.6 or 4.7 can be done with GPT 5.4, maybe on the margins. And we're talking
00:07:52about extreme complexity. You can argue for one or the other, and there will be tons of people who
00:07:56argue for the open AI models over the anthropic models, but for 99% of people, there's really not
00:08:02going to be a huge difference. And so when it's, Hey, should it be called code or should we be using
00:08:07codex? The answer is just pick one. What's your personal preference. Now, what about something
00:08:12like anti-gravity which plays the part of harness slash IDE kind of, kind of still in the space
00:08:18though. Where does this go? Well, not too long ago, I would have put it S tier and said there was kind
00:08:21of this triumvirate up here of these coding agents, but these days I would put it in A tier. The truth
00:08:27is anti-gravity when being used with the other Google models like Gemini 3.1 just isn't as good
00:08:32when it comes to cloud code with Opus or codex with something like 5.4, you'd be better off using
00:08:39anti-gravity with those other models, which then begs the question, why even use the anti-gravity
00:08:44shell around it to begin with? Why not just use something like cloud code or open AI with their
00:08:49native models that it's actually built for. Beyond that, you're starting to see some people complain
00:08:52about usage issues with anti-gravity as well. When, when it first came out, they pretty much gave you
00:08:58full usage for free forever it felt like. Now the next tool we're going to talk about isn't really a
00:09:03coding agent or harness yet it's often talked about with these others, so we're going to include it
00:09:07here and that is OpenClaw. Now I've been pretty vocal in the past about my dislike for OpenClaw
00:09:12and to be honest it hasn't changed. I would put this as C tier and arguably D tier. OpenClaw
00:09:18is performance theater. When you're using OpenClaw, it feels like a lot's happening. We're doing
00:09:23dashboards, we're connecting stuff, it has a heartbeat, it's talking to me. There's a lot of
00:09:28motion going on but there's not a lot of forward movement. You feel like you're doing stuff and for
00:09:32a lot of people that's enough but the fact is what four months later, three months later and this will
00:09:39still be the case three years from now. I'm still waiting for someone to come and tell me oh here's
00:09:44this OpenClaw use case I, I do that can't be done inside of cloud code or codex and can't be done
00:09:52more efficiently by cloud code or codex and if you tell me it's a morning report or some sort of
00:09:57brief I'm going to ban you from the channel. And this low ranking of OpenClaw only has become more
00:10:02obvious ever since Anthropic came out and said you can't use your max subscription inside of OpenClaw
00:10:10which means you need to pay API prices for something like Opus 4.7. Now caveat literally today something
00:10:16came out where they're like oh no uh we it's in the OpenClaw docs where it says we talked to someone
00:10:21on the staff at Anthropic and they said it's good now. I, I feel like we need a little more proof than
00:10:28some one-liner in their own documentation saying we talked to a guy. So until that's confirmed it's
00:10:36crazy to think that like OpenClaw is some sort of cloud code replacement and I get they're not the
00:10:40same thing yet people will use this over the other or they'll say oh no my OpenClaw calls cloud code
00:10:46to do stuff. Why? Why not just call cloud code yourself? And I'll kind of end this discussion here
00:10:52but OpenClaw and in some pretty much same sort of thing with Hermes right I think Hermes is a little
00:10:59better but I'm gonna put it solid C tier these are just productivity theater. Feels like we're doing
00:11:06a lot but you can do all this inside of an actual coding agent you just can. Now Hermes kind of has
00:11:12some clever things going on with you know these self-updating skills but the self-updating skills
00:11:18thing also gets very close to this illusion of something good happening where it's like oh no
00:11:25every 15 tool calls like it's taking a look at its skills and it's rewriting them like does it need to?
00:11:29How are we determining that this new one's actually working? Why wouldn't I just use cloud code and use
00:11:36something like the skill creator skill to take a look at the skills I've been using regularly or I
00:11:40don't know actually look at the outputs myself and determine if we need to do that. So these sort of
00:11:46like tools you could also throw something like paperclip in here a lot going on nothing happening.
00:11:53Now you'll notice I'm not covering some other popular coding harnesses things like
00:11:58open code the truth is I just haven't used them very much I don't feel like I can speak
00:12:01intelligently on them and that becomes I think a discussion that goes into the open source model
00:12:08route which I think is a whole other path forward. I think if you are deep in the open source open
00:12:15model you know type route you kind of already know what you're doing because the average person is not
00:12:21going to be touching that stuff nor do they have the hardware to even do it at any level. That being
00:12:25said I've heard nothing but good things about stuff like open code and pi and so if you like it keep
00:12:29doing it. Now let's move into some of these no code builders things like lovable in n8n and I get there
00:12:36is some delineation there between like automation forward stuff like n8n and lovable but we're going
00:12:40to evaluate them all under one umbrella and so first up is lovable lovable. If you ask me this
00:12:48question you know nine months ago lovable I would have had that S tier I love loved lovable when it
00:12:56came out it pretty much took that base of what was like what three point sonnet 3.5 at the time opus
00:13:02three at the time and really tuned it to give you great front-end design and even though it was kind
00:13:08of expensive it didn't seem crazy. Now lovable it's like what's the point what's the point it's
00:13:16lovable is lovable going to be around in a year two years when would you suggest someone to use lovable
00:13:27like there's no point there's no point when codex exists when clod code exists when even some of
00:13:33these chatbots exist you're paying an insane premium to have like a one-click super base
00:13:40setup or a one-click deploy and it these other systems have gotten so good especially stuff like
00:13:47clod code at creating front-end designs that are just as good as what lovable produces that it's
00:13:52like what are you putting the price on for convenience like how much are you willing to pay a
00:13:57month for convenience because it's an insane premium and in a certain point the convenience
00:14:01doesn't matter because if you are building a real app there is some sort of technical complexity or
00:14:06at least roadblocks you're going to have to navigate no matter what and if you could do that you could
00:14:11do it in these you could do it in anti-gravity so like lovable is not a bad product by any means but
00:14:17when you compare it to the competition and sort of the ecosystem at large it's tough to justify its
00:14:23existence i mean is anybody recommending lovable these days i'm certainly not and that applies to
00:14:29bolt as well which is just at this point another version of lovable remember when bolt first came
00:14:33out it was like wild like oh my gosh this thing's like creating an app from scratch and what are we
00:14:38a year and a half later i'm like ah d tier would never recommend it they're like this is a dead
00:14:43space and when i don't have on your stuff like base 44 and those types of things like replit
00:14:49like i think that's where they're going as well i haven't used them enough to give you great answer
00:14:53but again i think if when you look down the line and you see yourself as ultimately competing
00:14:59against clod code and codex and anti-gravity they're only gonna get more user-friendly i mean
00:15:05look at stuff like co-work like it's gonna be more user-friendly and you like the cultural zeitgeist
00:15:11here is clod clod clod clod like if you're repli if you're base 44 certainly if you're bolt or lovable
00:15:17what chance do you have in the long term i don't think it looks great now cursor is an interesting
00:15:22one because it can be used in combination with clod code and codex and i did that for a long
00:15:27time i essentially used cursor as my ide as my vs code on steroids and use of a clod code
00:15:34and the idea was okay that clod code is going to do 99 of the work but if i ever want to cross-check
00:15:39the work well i'm inside of cursor so you know i could have a gpt 5.4 take a look directly at the
00:15:46work clod code did and say hey that was good that was bad it was a very easy to use second set of eyes
00:15:52these days i what do we need that well first of all you certainly don't because i can use codex inside
00:16:02of clod code there's a plugin that we can use and we can have a do code review adversarial code review
00:16:08so that particular use case is out the window do we love composer 2.0 me maybe you do if that's the
00:16:15case and you love composer maybe it's a tier maybe it's kind of like in the anti-gravity space but for
00:16:21me these days i'd say it's b-tier like it's not a bad product at all don't get me wrong i'm not saying
00:16:25you shouldn't use cursor i think it's a great product it's just again like who's the cursor
00:16:30customer and why are they choosing cursor over these other guys i i think it's kind of one step
00:16:37away from that whole like lovable problem and i probably should have talked about these guys
00:16:41in the last section but gemini cli name 10 people who use gemini cli who don't work for google
00:16:47got it i think i by virtue of literally being absent from the discussion and again does anybody
00:16:55actually use this if the answer is no there's probably a reason for that i i literally do not
00:16:59know a single person who uses gemini cli or has stated they use it who quite literally isn't a
00:17:04google employee and i think that speaks volumes now how about n8n oh i was a super n8n fanboy for
00:17:11a long time so this was an automatic s tier for a long long time these days n8n went from this is
00:17:18the go-to tool you should use to create ai agents and automations without using code to a very niche
00:17:26tool because let's be honest everything n8n can do claud code can do everything n8n can do codex can
00:17:33do and if you are a tool where i can say everything you do claud code or codex can do there better be a
00:17:40very compelling reason or compelling differentiator whether that's ease of use cost niche to justify
00:17:48your existence for n8n that still has that and i think it's niche and ease of use but i so i only
00:17:55use n8n these days for client projects if it's a client project for a very non-technical team where
00:18:01for whatever reason they also need to be hands-on so they at some point are going to need the ability
00:18:06to go inside the automation and maybe tweak one or two things n8n makes that very easy to do the
00:18:11visual setup is great but i mean again the the ease of use now is becoming n8n's differentiator it isn't
00:18:19because it's like way faster either than claud code like i can spin up i can spin up and i can spin up
00:18:27n8n automations faster inside of claud code then i can do so through n8n itself that's kind of a
00:18:32problem for long-term viability i think i mean so it went from i think kind of ubiquitous in this
00:18:39no code ai space to like maybe if you need it for a specific reason and you need some justification
00:18:45to do so but like the product itself hasn't changed it still does what it does very well but n8n's
00:18:50direct competition like make in zap year d tier more expensive less flexibility why bother so again
00:19:00what is your differentiator if you can't do that you're gonna die so i don't know why anyone would
00:19:04use maker zap year at this point beyond legacy reasons and let's bring in some of the stragglers
00:19:09at this point notebook lm where does notebook lm fall it's kind of just like a niche ai tool like
00:19:15it's not a coding agent it's not really a no code thing although there's an argument to be made that
00:19:20it is right if i can give it a prompt and give it context it creates things for me so sort of
00:19:25that being said notebook lm is amazing and it is s tier for sure for like a one-off ai tool i think
00:19:33notebook lm is the best out there i think it's the best google product to be frankly honest and the
00:19:38fact that i can integrate it inside of claud code certainly can integrate it with anti-gravity using
00:19:43stuff like the notebook lm pi cli it's just such a giant productivity boost and google lets you do
00:19:50so much on there for free which is wild and you can create really good deliverables things like
00:19:55slide decks infographics and it's the best way in my opinion for doing any sort of research that
00:20:01involves lots of youtube videos so i've done a ton of content on combining claud code and notebook lm
00:20:07for great reason because notebook lm is amazing it's free another one we didn't talk about maybe we
00:20:13should have in the chatbots is perplexity now perplexity is a chatbot yet is anyone using
00:20:20perplexity as their only standalone chatbot maybe maybe the argument these days with perplexity is
00:20:26it's the chatbot that lets you use a bunch of other chatbots but for a long time it was s tier for me
00:20:34and again like so many tools it's just starting to fall down these days i would put perplexity like c
00:20:40tier and maybe that's in large part because i'm not a user of perplexity's like periphery tools like
00:20:47computer or comet their browser and so because of that i find myself really only using perplexity
00:20:53and i'm on the pro plan i really use perplexity when i'm like all right i just want to get a
00:20:58information about something that happened like today and i want the sources and i don't want to have to
00:21:03like overtly prompt claud or chat gpt to do a web search to get me that like real data so in a sense
00:21:10it's almost like grok on steroids where it is really good at web fetching and it it's built to
00:21:16do that but again daily driver no is it worth paying 20 bucks a month on top of these other
00:21:22tools no it's like a nice to have it's a nice to have and in our last category we're going to talk
00:21:28about some of these content creation tools image generators video generators that sort of thing now
00:21:33first on the list is a recent product and that is design by claud now in a vacuum i would put this
00:21:42s tier i absolutely love design i've you know been using it a ton these last few days and it is a huge
00:21:49step up from baseline claud code and it's also a huge step up from most of the skills you see out
00:21:55there things like well certainly the front end design skill the ui ux pro skill blows it out of
00:22:00the water and right off the rip it just does a great great job at allowing you to very quickly
00:22:05iterate through a bunch of visuals however comma the usage is the worst thing i've ever seen in my
00:22:12life you burn through it at a crazy rate i don't get extra usage even though i'm paying 200 bucks
00:22:18a month i have the same usage as someone on pro what why and it's not really clear at all what
00:22:23eats it all up although once you do some experimentation after you burn through it all
00:22:26becomes very obvious doing a design system can be 30 of your tokens asking for a simple landing page
00:22:32can take up like five percent of your tokens so the usage is tough in a vacuum s tier d tier usage
00:22:39so i think at the end of the day of b tier will kind of just split the difference great tool once
00:22:50they have the usage will be amazing one of the best products to be honest but right now it's not really
00:22:55usable for the average person i'm already using extra usage for it oh yeah copilot i even mentioned
00:23:00copilot i forgot this thing was even in here probably speaks for itself copilot you use it when
00:23:06you're forced to nobody's using this on their own there's some arguments to be made that it's decent
00:23:11at like all right like it's integrated into the company itself so um the c tier i get c minus tier
00:23:21i don't know the fact i even forgot to mention it probably says a lot now clinging clinging
00:23:263.0 if you don't know what cling 3.0 is it is a video model and it is solid a tier really really
00:23:33good stuff its direct competition is vo3 well via 3.1 these days which i would put it b tier it's
00:23:40kind of getting outdated at this point and it's very expensive which is the other thing um and
00:23:47then we have seed dance 2.0 if you've seen any ai video lately it's been seed dance remember all
00:23:53those videos like a month ago when it was like embedded beta of all the things like actors and
00:23:59movie scenes being recreated that was seed dance and seed dance is great definitely a step up from
00:24:04vo and cling the the visuals are wild it's wild and surprisingly not that expensive comparatively
00:24:11and then we have our image models nano banana pro is s tier it's been best in class and best in class
00:24:20for a while and then we also have the brand new image gen 2 from gpt now from the little i played
00:24:28with it and this like just came out i it would be s tier but i haven't used it enough and seen it
00:24:34enough edge cases so this is one of those scenarios where like you know if you're bullish on it s tier
00:24:42if you want to wait and see it's at least a tier um so we'll keep it in a tier from now we'll see
00:24:49how it holds up um but that being said a great image model is a big boon for chat gpt because
00:24:56tons of people use this tons of people use image gen and one of the big differentiators for gemini
00:25:01versus chat gpt has always been its superior image models so here is where we landed i hope this sheds
00:25:09some light on what you should be focusing on and what really isn't worth your time and hopefully it
00:25:15even brought up some new ai tools that you've never even heard of and might want to check out the big
00:25:20players i don't think you should be surprised things like clod code and codex have been dominating
00:25:24for a while but more importantly i would say stay away from stuff in the c tier below you know with
00:25:30certain exceptions like we talked about when it comes to clod's chatbot i think you can very easily
00:25:36get lost in the weeds and lose a ton of money especially on these things like hermes and open
00:25:40claw let alone stuff like lovable invest your time in these a tier and s tier tools that will actually
00:25:47give you a return and i think that really goes for clod code and codex and anti-gravity too in that
00:25:53respect and that if you get proficient at these tools you got proficient at them because you became
00:25:59proficient at the fundamentals of building with ai so if next week clod code falls off a cliff and
00:26:05codex falls off a cliff and some other harness becomes you know s tier and the one everyone's
00:26:10using well those are transferable skills those are transferable skills that aren't tied to a specific
00:26:15platform in a specific tool set you've been learning how to build you've been learning the fundamentals
00:26:20of software architecture and that's important and that's ultimately what you should be driving for
00:26:25so hope you liked the video let me know what you thought as always check out chase ai plus
00:26:31if you want to get your hands on my clod code masterclass and i'll see you around