40 AI Tools Ranked from Essential to Forgettable for 2025
An honest review of which AI tools are actually worth your time
If you're looking for new AI tools to try, this post is for you.
The AI landscape is moving fast—new models are emerging daily, some tools are leveling up, and others are falling behind. I've ranked 40 of the most popular AI tools based on how useful they actually are in 2025. This list has grown since my previous roundups, thanks to the sheer number of innovative tools hitting the market.
Some of these rankings will be controversial, so I'll explain where each tool shines and where it falls short. Let's start at the bottom and work our way up to the must-haves.
D Tier: Not Useful or Fun
Let’s kick things off with four disappointing picks:
Apple had the chance to redefine personal AI by replacing Siri with an above-average LLM, bringing AI editing features into Photos, and integrating AI agents with each iOS app, but so far, it’s been a letdown. The early reviews were bad, and even months later, not much has changed. The message summaries miss the point, AI features are hit-or-miss, and it doesn’t compete with what OpenAI and Anthropic are doing. If Apple wants to stay in the AI race, they need a major revamp.
Adobe Firefly promised to shake up the generative AI scene with its ethically trained models and integration into Creative Cloud tools like Photoshop and Illustrator. I gave it a shot, expecting a smooth image and video generation experience, but the results have been underwhelming. The images often look like stock photography—lacking the creativity or detail of Midjourney or Flux—and the video features, still in beta, feel clunky and limited. Sure, it’s “commercially safe” with its licensed training data, but that safety comes at the cost of quality. It’s a nice idea for pros already in the Adobe ecosystem, but for now, it’s not practical enough for my workflows, landing it in D-tier.
Poe gives you access to top AI models (GPT-4, Claude, etc.) for one $20 subscription. Sounds great, right? But in practice, it’s just okay at everything. The usage is limited, the UX is clunky, and it’s missing features like ChatGPT’s voice input, Claude’s advanced formatting, web search, deep research, etc. It feels like a bargain bin bundle, not a value driver.
I tried Grammarly for a week, but its generative AI suggestions often miss the mark, changing the tone or even the meaning of my sentences. It’s a nice idea—automatic rewrites for clarity or brevity—but I end up undoing a lot of its edits. Maybe I just didn't use it enough for it to understand my tone, voice, and writing style. I may return, but if they can nail this out-the-box, Grammarly could be A-tier.
C Tier: Promising but Not Daily Drivers
These tools are doing interesting things, but they’re not consistently providing daily value yet:
NotebookLM is best for turning text into podcasts. It can generate fun 10-minute podcast episodes, but I don’t use it much beyond entertainment, so it’s C-tier.
Google hyped Gemini as a GPT-4 rival, but in my testing, it’s consistently behind ChatGPT and Claude. Despite its huge context window and multimodal capabilities, responses often feel too brief or miss the nuance. It's just not competitive with leading LLMs to warrant any daily use, though I'll keep my eye on it in 2025 in case this changes.
Suno is a fun tool for creating AI-generated songs (it even works in multiple languages), and it feels magical the first time you use it. But beyond entertainment, I haven’t found a practical use case, so it’s D-tier for me. It’s fun, but I use AI tools for work. Musicians may get value out of this tool however, as well as those who use a lot of music in their daily work.
Sora produces the most realistic AI-generated videos. Its strength is realism—videos can look uncannily real (though longer sequences still have some weird physics or continuity issues). OpenAI’s “Sora Turbo” is faster and better than early previews, and it’s integrated into ChatGPT. It’s limited (short durations) and requires a ChatGPT Plus/Pro subscription, but is a top AI video generator. It's not the clear leader however. Google's Veo 2 is just as good, and open source options like Wan and Hunyuan offer similar quality but with more control. Sora needs more features and more accessibility to replace any of these tools.
Manus represents a new wave of autonomous AI agents that can figure out how to do tasks for you. You can ask it to perform complex, multi-step jobs—like web research or data analysis—and it will plan and execute them independently. In early trials, it’s topped some AI agent benchmarks, and when it works, it feels magical. But it’s still in closed beta, hard to access, and prone to errors, so it’s B-tier for now. Note: it also relies on an open source repo called Browser Use, which let's you do a lot of the same things, though coding knowledge is needed.
ComfyUI is an open-source, node-based tool for running image and video models locally. It’s built for AI enthusiasts, letting you experiment with cutting-edge community models like FLUX (a new image generator that can outperform Midjourney with a lot of modifications) or Wan 2.1 (a top open-source text-to-video model). It requires technical setup, so it’s not ranked as a standalone product, but it’s worth exploring if you want to run AI on your own hardware. Note: it's notoriously hard to use and has a steep learning curve, so much so that I can't use this as a daily driver without going crazy.
Playground AI is best for logo and social media design. It has a great image-to-image feature, and allows you to generate multiple variations, including different color variants. I wish it was better at creating designs with consistent branding (they’re working on it), and had more control (like Canva) but it’s still a solid tool. It’s B-tier because it’s early and not essential for me daily, but I see potential in it taking over some of the things Canva is good at.
Pika is a newer player in AI video generation focused on creative use-cases. In the past few months, it's stopped competing head-to-head with other AI video generators and is instead investing in features like inserting characters into videos, special effects, and creating videos using ingredients. Some of the results it generates are mind-blowing. It’s great for creative projects and standing out on social, but its features are more of a nice-to-have than something you can rely on for work. This may change if its quality can become indistinguishable from film-level visual effects.
B Tier: Best-in-Class, But Not Essential, Yet
These tools are the best at what they do—often leading their niche—but I don’t find them essential for my daily work. They’re the specialists I turn to occasionally or just for fun:
Kling offers some of the best AI-generated video results at a budget-friendly price. It can create cinematic-style videos, lip-sync animations, and even camera motion effects. If you want an AI video generator without spending OpenAI Sora money, Kling is your best bet.
Magic Hour is an all-in-one platform for image and video creation. It offers a suite of generative tools—video-to-video, face swap, image-to-video, text-to-video, and more, in a web app. Its standout feature is that it offers 90% of its products for free. I've used its tools daily for when I need to get an editing task done fast. Disclaimer: I do video/image editing every day, and I'm one of the co-founders of Magic Hour, but I'd genuinely still recommend it if I wasn't.
Runway’s generative video and editing tools are some of the best on the market. It lets you generate short clips in various styles from a text, image, or existing videos. I’ve used it to prototype video ideas, and it delivers impressive results without needing filming or animation skills. Many reviewers say it beats OpenAI’s Sora in quality, and it's close. Runway is in striking zone from taking a leap upward from other video generation players—it's awesome, but so far more useful when you're working on a complete video project.
HeyGen is best for creating talking head videos, and is tied to Synthesia for quality. Some AI creators are using it to grow their social channels by scaling the number of talking head videos they can do, but the videos still have an uncanny valley effect. I've seen friends use it, and it doesn't feel like the real them—it's souless. It’s great for what it does, but not something I'd rely on personally, so it’s B-tier.
Gumloop is a no-code AI automation framework that allows users to design automated processes with a drag-and-drop interface and integrate with third-party apps. It's versatile, enabling tasks like document processing, web scraping, and SEO automation, which are helpful for activities like marketing. However, it comes with a steep learning curve, so I only rely on the workflows others create, and its advanced features may not be necessary for simpler workflows. I think if Gumloop can solve for discovery, and even use AI to create workflows personalized to each user's business, it has the potential to be more valuable than Zapier.
Granola is best for meeting notes, which is a crowded space with a lot of commodity products. It could be a huge time-saver for my team, but we're just a team of 2 and don't do a lot of external meetings. It's also more valuable as more and more companies adopt it. It'll be interesting to see if Granola can maintain the lead they have, because its benefits really accrue at scale due to data advantages and network effects.
Google Gemini AI Studio (Image Editor)
Google’s new AI Studio, powered by Gemini Flash, brings powerful multimodal editing tools. You can upload an image and edit it through natural language prompts in a conversational way—telling Gemini what changes you want, step by step. It leverages Google’s world knowledge to keep edits consistent. It’s a promising tool for users who want to iterate on images without manual Photoshop work. But it’s still experimental, and output quality can be very hit-or-miss, so it’s B-tier for now.
PhotoRoom uses AI to remove backgrounds and enhance photos, and it’s popular for e-commerce. It's blazing fast and higher quality than remove.bg and Apple's background removers. It's also adding smart AI editing features that give you suggestions on editing steps. It’s a solid tool, but I don’t edit photos enough to make it a daily necessity, and there are so many free alternatives that I can't justify paying for it. Photoroom is best for those do photo editing at scale and professionals who need that 5% quality improvement.
OpusClip uses AI to turn long videos into short, shareable clips with auto-captions and highlights. It’s a time saver for content creators who need to repurpose podcasts or webinars into bite-sized social media content. I’ve used it to clip a podcast into TikTok-ready snippets, and the auto-editing is spot-on. But since I don’t edit videos daily, OpusClip lands in B-tier—excellent for its purpose, but not something I use all the time.
A Tier: Best-in-Class and Part of My Daily Workflow
These tools are not just the best in their category—they’re tools I use often and rely on for work:
For casual coding, Cursor is still the best AI tool. It makes writing, debugging, and improving code way easier, even if you’re not a full-time engineer. It’s like an IDE with an AI pair-programmer baked in. I can start projects just by entering a prompt, and talking to it. I can also use it like autocomplete by starting off a coding task and asking Cursor to complete or debug it.
Windsurf, by Codeium, recently upgraded to Claude 3.7 Sonnet, making it even better at coding than before, and arguably, better than Cursor. It offers features like tab-completion and tool suggestions, and it’s catching up to Cursor fast post 3.7. I’ve tried Windsurf, and it’s impressive—especially since it’s free for individuals—but I keep coming back to Cursor for its interface and flow. Still, Windsurf is A-tier for its utility in coding workflows and it's neck-and-neck in this race.
Perplexity does search well. It’s an AI-powered search engine that gives direct answers with cited sources, which is the killer feature. Whether I’m researching topics, it delivers accurate results I can verify. I use it daily, but it's starting to take a backseat as ChatGPT search begins to replace that role for me. ChatGPT search is quite good, and it's in a tool I already use a lot. Perplexity's new features like Deep Research just aren't that competitive however. That said, Perplexity still deserves an A-tier spot.
ElevenLabs is my go-to for text-to-speech and voice cloning. It generates natural, human-like speech in various voices and accents. I’ve used it to create tutorial voiceovers that sound like a real person. It supports many languages and has emotional intonations that most TTS tools can’t match. Sesame recently launched and its open source model is showing promise for besting ElevenLabs with some fine-tuning, so it remains to be seen if ElevenLabs can maintain its lead. It’s also a bit niche for my workflow—I don’t need voiceovers daily—which is why it’s B-tier, but it's still best-in-class.
If you need to go from a design idea to a working prototype, v0 is incredible. It generates interactive mockups with good spacing. I use it regularly as one of three products for rapid prototyping of landing pages and product flows. It also takes in images, allowing you to remix existing designs. When I have an app idea or feature to spec out, v0 is one of the first tools I open. It falls short in aesthetics. 90% of its generations are black and white and extremely basic, so I only rely on it for structure rather than finished assets.
Lovable is an AI-powered app builder that turns ideas into working web applications. Think of it as your cracked full-stack engineer intern on caffeine. You describe what you want (“a meme image generator”), and Lovable generates a functional prototype—complete with UI and working code. It’s incredibly empowering for non-developers, and I always open it with v0 when I'm mocking designs. It does fall a bit short in quality however. Some of the designs are pretty corny, like a high-school project, and it takes a lot of prompting to get it right. I keep it in A-tier because I don’t need to spin up prototypes daily, but for prototyping, it’s fantastic.
Magic Patterns is a designer’s version of v0 and Lovable. It uses generative AI to create functional UI components and design mockups. What sets it apart is its export to Figma feature, which is a life saver because it allows you to get quality components fast and iterate on designs with a fine touch. I’ve found it great for remixing my designs and iterating collaboratively. The only problem is it's kind of buggy compared to other tools - some of the components don't transfer to Figma, or never finish rendering, but when I need design help, it’s a huge time-saver and great pair-designer.
Midjourney is still the gold standard for AI image generation. It generates stunning, detailed visuals that stand out and is great for art, design concepts, or illustrations. It fits well with image-to-video tools and is what most AI creators use on social media. When I need the absolute best image output, Midjourney is my go-to, and some of the creators who use Magic Hour use it as a part of their video workflow, and they do really well on social media. It's S-tier quality, just not S-tier frequency for me.
S Tier: AI Tools I Can't Work Without
These are the AI tools I rely on as daily essentials, and if they went down I'd probably lose half of my productivity. They're also the usual suspects. Here's a breakdown of what they’re best at:
Grok, built by xAI, is 100% free and has no limits I can perceive as an X subscriber. It's great for simple image generation, coding, and producing writing that sounds human. It's also integrated with X meaning I can read tweets unlike most LLMs. It's great for getting to the point and talking to you like a human. I've used it to talk through decisions and get its feedback, and I trust it more than other tools because it tells it like it is. The quality and consistency just doesn’t outmatch tools like ChatGPT, which has a richer feature set, and Midjourney, which is superior for image generation, but when you need something unconstrained and raw, Grok is king.
Anthropic’s Claude has become my writing and coding go-to. Sonnet has a human-like voice that has more personality than ChatGPT 4o. It's also a meticulous editor and long-form writer that generates copy that humans prefer over ChatGPT. I use it to:
- Review and improv my writing
- Handle long context inputs (it has a 100k token context window)
- Complete coding tasks that ChatGPT 4o struggles with
For instance, I’ll say, “Here’s a draft of my blog post, rewrite it so it's more clear and concise” and it provides an edited version that elevates clarity but keeps my tone. With the recent Claude 3.7 Sonnet model, its reasoning and coding skills got even better, closing the gap with ChatGPT in many areas. But it still lacks basic image and voice capabilities, and the usage limits are so low, even for paid users, that it's just not as useful for back-and-forth iterations as ChatGPT and Grok. I use Claude and ChatGPT in tandem daily—but for certain writing and coding tasks, I turn to Claude first.
The original breakout AI is still an everyday necessity. ChatGPT (with GPT-4o) is like an extension of my brain. I use it for:
- Writing—Copy, emails, creative content, etc., ChatGPT 4.5 is slightly better than 4o, but even 4o is solid. It's sometimes bested by Claude or Grok, so I often compare between them.
- Coding—I still use it daily to fix my bugs and generate new code using o3-mini.
- Research—Deep Research is the most in-depth research tool out there. It's well-sourced, and can be cut down into informative, impactful bites.
ChatGPT has multimodal capabilities (it can see uploaded images and speak/listen via voice) and an ecosystem of plugins for specialized tasks, including using a computer through Operator (which I'll admit isn't quite game-ready). OpenAI also offers the O1 model for solving complex problems and Sora for video generation. Paying $20/month for all of this is a no-brainer, and I even paid $200 for a month to try Deep Research and Operator. If I could only pick one AI tool, I’d pick ChatGPT.
Final Rankings
Here’s the full tier list again, summarized:
D Tier: Not Useful or Fun
These tools failed to deliver on expectations—either underwhelming, impractical, or easily replaced by better options.
- Apple Intelligence – Siri’s AI glow-up still falls short of OpenAI and Anthropic.
- Adobe Firefly – Feels like stock photography in AI form. Limited, bland, and outclassed by competitors.
- Poe by Quora – A bundled AI subscription that’s just okay at everything but great at nothing.
- Grammarly – AI writing suggestions that often make text worse instead of better.
C Tier: Promising but Not Daily Drivers
These tools have interesting use cases but aren’t reliable or valuable enough to use daily.
- NotebookLM – Good for turning text into podcasts, but not much else.
- Gemini (Google’s AI Model) – Overhyped, still behind ChatGPT and Claude.
- Suno – Fun for generating songs, but not useful unless you work with music regularly.
- OpenAI Sora – Impressive realism, but short videos and limited access hold it back.
- Manus – AI agents that automate complex tasks—magical when they work, but still unreliable.
- ComfyUI – Powerful for running AI models locally, but brutally difficult to use.
- Playground AI – Good for quick social media designs, but lacks the control of pro tools.
- Pika – AI video generation with cool effects, but still more of a novelty than a work tool.
B Tier: Best-in-Class, But Not Essential (Yet)
These tools lead their niche but aren’t critical to my daily workflow.
- Kling – High-quality AI-generated video, great for cinematic effects.
- Magic Hour – Versatile AI-powered video and image tools
- Runway – A strong AI video editor, especially for creative projects.
- HeyGen – Best-in-class for AI talking head videos, but still feels unnatural.
- Gumloop – A promising no-code AI automation tool with a steep learning curve.
- Granola – Solid for meeting notes but best for teams at scale.
- Google Gemini AI Studio (Image Editor) – AI-powered image editing with potential, but still inconsistent.
- Photoroom – The best background remover for e-commerce and product photography.
- OpusClip – Auto-edits long videos into short, social-friendly clips.
A Tier: Best-in-Class and Part of My Workflow
These are the best tools in their category. I don’t use them daily, but when I do, they’re essential.
- Cursor – My go-to AI coding assistant.
- Windsurf – A strong alternative to Cursor, powered by Claude 3.7.
- Perplexity – AI-powered search with cited sources.
- ElevenLabs – The best text-to-speech and voice cloning tool available.
- v0 – AI-powered prototyping for UI and product design.
- Lovable – An AI app builder that quickly turns ideas into working prototypes.
- Magic Patterns – Great for AI-generated UI components, with Figma export.
- Midjourney – Still the best AI image generator for high-quality, stylized visuals.
S Tier: AI Tools I Can’t Work Without
These tools are essential. If they went down, my workflow would take a serious hit.
- Grok – Free, no-limit AI chatbot with X integration, great for raw, direct conversations.
- Claude – The best AI for thoughtful writing, editing, and long-context reasoning.
- ChatGPT – The most versatile AI assistant for writing, coding, and research.
That’s my full tier list of the best AI tools in 2025. What do you think? Let me know if you agree or if I missed any must-have tools.
.png&w=3840&q=100)
.png&w=3840&q=100)
.jpg&w=3840&q=100)
.png&w=3840&q=100)