My Go-To AI Tools (Archive)
EDIT: The first “Go-To AI Tools” post came out on October 17, 2024. This archive contains all prior versions. The newest version is always found here.
I often get asked about AI tools that I personally use.
While I regularly review all sorts of AI models and sites, most of them don’t end up becoming a part of my routine.
Here are those that did.
February 2025 snapshot:
🗣️ 1. Large language models
These are the chatbots and LLMs I turn to.
🎁Free:
Claude 3.5 Sonnet for brainstorming and day-to-day usage. Claude has a “fresh” conversational feel and tends to come up with unconventional ideas when brainstorming. The Artifacts feature is fantastic for working on a project side-by-side with Claude. The only downside is the strict message limit for free accounts. (Update: I find myself using Claude much less after I renewed my ChatGPT Plus subscription and DeepSeek-R1 came out.)
DeepSeek-R1 for reasoning tasks and a surprisingly fresh, natural tone of voice when coming up with creative ideas. It also runs HTML code directly in the interface, letting you test apps and iterate (similar to Claiude’s Artifacts). (Update: New addition.)
Gemini 2.0 Pro Experimental (via Google AI Studio) for working with long documents or files. Gemini’s insane 2M-token window and ability to natively process audio and video make it a beast for dealing with any long inputs in multiple formats. It’s also my default “beta reader” when asking for writing feedback. (Update: This replaced Gemini 1.5 Pro in my original post.)
GPT-4o (via ChatGPT) for mixed media conversations. GPT-4o is my least preferred LLM in terms of tone of voice and conversational ability. But it’s hard to beat the complete free ChatGPT package with web browsing, (limited) image creation, and more. ChatGPT is also my go-to partner for working on AI Jest Daily cartoons.
💵Paid:
ChatGPT Plus. It’s a true all-in-one solution that gives you…
Advanced data analysis (“Code Interpreter”)
Image creation via DALL-E 3
Web browsing for news and knowledge beyond its training data
Ability to create your own Custom GPTs for different purposes
Canvas: “Artifacts”-inspired right-side window for working on coding or writing projects.
Advanced Voice Mode: Extremely humanlike, real-time voice chat.
Reasoning models like o1 and o3-mini for advanced reasoning and coding tasks.
Deep Research (upcoming): The most impressive research agent to date. Available only to Pro accounts for now, but coming to Plus accounts soon.
For $20 / month, this set of features is a steal.
🖼️ 2. AI images
Here are my favorite image models.
🎁Free:
DALL-E 3 for cartoons and simple illustrations. DALL-E 3 was my original choice for single-frame cartoons for AI Jest Daily. It still holds up for that purpose, but for most other images I now turn to Imagen 3 (see below). You can use DALL-E 3 for free via Microsoft Designer or Microsoft Copilot.
Imagen 3 (via Google Labs) for most of your image needs. A truly solid, prompt-adherent model that can spell text and create great images for free. (Update: New addition.)
Ideogram 2.0 as a solid all-rounder model. It does better at photographic1 images than DALL-E 3 and is one of the better image models at handling complex prompts and accurately rendering long text inside images. You get 10 generations (40 images) for free every day. (Update: Mostly replaced by Imagen 3.)
FLUX 1.1 [pro] if you want photographic images. I can’t say I’ve used FLUX nearly as much as the other models, but it’s excellent at rendering realistic “photos.” Two sites let you use FLUX for free. (Update: Mostly replaced by Imagen 3.)
💵Paid:
Midjourney…I know, big surprise! I’ve written dozens of posts about Midjourney, and it remains the best paid option in my book. It can handle a huge range of styles and image formats, gives you strong editing tools and features, and is still the best model for realistic photographic images.
📽️ 3. AI video
Here’s my take on AI video models after testing lots of text-to-video and image-to-video options.
🎁Free:
Kling AI for text-to-video. Kling AI comes close to Runway’s quality while giving you free daily credits. Kling AI also recently upgraded its lip-syncing feature and added a text-to-speech engine with realistic voices.
Hailuo AI for image-to-video. It came seemingly out of nowhere to truly impress me with its visual consistency and realistic movements.
💵 Paid:
🎵 4. AI music
I rarely use AI music tools (and mostly just to goof around), but when I do…
🎁Free:
Suno and Udio. The two are largely interchangeable for casual users, with similar feature sets and output quality.3 Both give you a generous amount of free credits, so take them for a spin to see which one you like the most.
Riffusion FUZZ recently entered the mix and does pretty well in my limited tests. (Update: New addition.)
💵 Paid:
Suno or Udio. Once you’ve picked your favorite, you can always upgrade for additional features. Suno might have a slight edge with its cool “Covers” option that makes a cover version of any song in whatever style you wish. Udio is slightly more robust in terms of vocals. You honestly can’t go wrong with either one.
🔬 5. Research
🎁Free:
NotebookLM is such an easy recommendation! I’ve been a fan since March 2024, and NotebookLM has only gotten better after that. You can now use audio and video as inputs and turn your sources into a range of helpful outputs including the incredible “Audio Overview” podcast. It can even help you cross-check your work against any sources you upload. NotebookLM is completely free, so you have no excuse not to try it for yourself.
Learn About is another Google product that complements NotebookLM nicely. While NotebookLM is grounded in your preselected sources and is built for synthesizing info across them, Learn About can browse the web and is built for open-ended exploration of any topic. (Update: New addition.)
Perplexity has been on my radar since early 2023, but it never became a regular tool in my arsenal. I still use it rather infrequently, but I find that its “Pro” search is a good midway point between simple web browsing and the thorough but time-intensive “Deep Research” tools. I turn to Pro search when I need a more robust answer than a simple search can provide but don’t want to wait 20 minutes for a long deep-dive report. (Update: New addition.)
Genspark has a poor man’s version of “Deep Research” which is great if you want to crawl hundreds of sources and generate a comprehensive report but can’t afford Google’s or OpenAI’s “Deep Research” products. My invite link gives you 1 FREE month of Genspark Plus. (Update: New addition.)
April 2025 snapshot:
🗣️ 1. Large language models
These are the chatbots and LLMs I turn to.
🎁Free:
Claude 3.7 Sonnet [NEW - replacing 3.5] for brainstorming and creative work. Claude has a “fresh” conversational feel and tends to come up with unconventional ideas. The Artifacts feature is fantastic for working on a project side-by-side with Claude. The only downside is the strict message limit for free accounts.
Gemini 2.5 Pro Experimental (via Google AI Studio) [NEW - replacing 2.0] for working with big documents or files. Gemini’s massive token window and ability to natively process audio and video make it a beast for dealing with long inputs in multiple formats. It’s also my default “beta reader” when asking for draft feedback.
GPT-4o (via ChatGPT) for day-to-day chats. GPT-4o has improved noticeably in conversational “vibes” over the past few months. Plus, it’s hard to beat the complete free ChatGPT package with web browsing, voice, (limited) image creation, and more.
💵Paid:
ChatGPT Plus. It’s a true all-in-one solution that gives you…
Advanced Voice Mode: Lifelike real-time voice chat with the ability to share your screen and camera for context-aware, multimodal conversations
Advanced data analysis (“Code Interpreter”)
Deep Research: One of the best research agents on the market
Reasoning models like o1 and o3-mini for advanced reasoning and coding tasks
Canvas: “Artifacts”-inspired right-side window for working on coding or writing projects
Web browsing for news and knowledge beyond its training data
Ability to create your own Custom GPTs for different purposes
For $20 / month, this set of features is a steal.
🖼️ 2. AI images
Here are my favorite image models. (This section was the biggest shake-up, with many models removed and replaced.)
🎁Free:
Imagen 3 (via Google Labs) for most of your image needs. A great prompt-adherent model that creates high-quality images at no cost.
Ideogram 3.0 [NEW - replacing 2.0] as a solid all-rounder option. One of the better models at handling complex prompts and accurately rendering longer text sequences inside images. You get 10 generations (40 images) for free every week.
💵Paid:
GPT-4o native image generation [NEW]: A true paradigm shift in AI images. Understands context, can render entire pages of text inside an image, handles consistent styles and characters, and more. (Included with ChatGPT Plus.)
Midjourney V7 [NEW - replacing V6.1]: Midjourney has been my go-to tool for years, and I’ve written dozens of posts about it. GPT-4o native image generation is now better at text rendering, scene complexity, and prompt following. But Midjourney is still great for exploring different styles, developing a personalized aesthetic, and creating realistic photographic images.
📽️ 3. AI video
Here’s my take on AI video models after testing lots of text-to-video and image-to-video options.
🎁Free:
Kling 1.6 for text-to-video. It often comes close to Runway’s quality while giving you free daily credits. Kling AI recently upgraded the platform and introduced cool features like “Elements” for blending many reference images in a scene. (EDIT April 16: Kling just released a massive 2.0 upgrade that might make it the best video model in the world. I haven’t had the chance to test this yet.)
Hailuo AI for image-to-video. It came seemingly out of nowhere to truly impress me with its visual consistency and realistic movements.
Wan 2.1 [NEW] is currently the best open-source video model. If you have the know-how and the hardware, you can even run it locally. If not, wan.video gives you free daily credits.
💵 Paid:
Krea [NEW] is excellent for sampling the best options in a single place. It pulls together many third-party video models (plus image and 3D ones). Krea also has a fun real-time image editor for iterative experimentation.
Runway Gen-4 [NEW - replacing Gen-3] is a fantastic recent model. The Runway site itself is a proper all-in-one creator platform that gives filmmakers dozens of robust tools to work professionally with the resulting video clips.
Veo 2 (Google) [NEW] is the best video model, capable of incredibly lifelike, consistent, and prompt-adherent generations. Unfortunately, it’s also prohibitively expensive at around $0.50 per 1 second of video—but you can use your Krea credits on it (see above). (EDIT April 16: Right now, Veo 2 is free to try in Google AI Studio with a few generations per day. Go go go!)
🎵 4. AI music
I rarely use AI music tools—and mostly just to goof around—but when I do…
🎁Free:
The three are largely interchangeable for casual users, with similar features, interfaces, and output quality.1 All of them give you a generous amount of free credits, so take them for a spin to see which one you prefer.
💵 Paid:
Any of the above: Once you’ve picked your favorite, you can always upgrade for additional features.
🔬 5. Research
🎁Free:
NotebookLM is such an easy recommendation! I’ve been a fan since March 2024, and NotebookLM has only gotten better after that. You can now use audio and video as inputs and turn your sources into different helpful outputs, including the incredible “Audio Overview” podcast. It can even help you cross-check your work against the sources you upload. NotebookLM is completely free, so you have no excuse not to try it for yourself.
Learn About is another Google product that complements NotebookLM nicely. While NotebookLM is grounded in your preselected sources and synthesizes info across them, Learn About can browse the web and is built for open-ended exploration of any topic.
Perplexity has been on my radar since early 2023, but it never became a regular tool in my arsenal. I still use it rather infrequently, but I find that its “Pro” search is a good midway point between simple web browsing and the thorough but time-intensive “Deep Research” tools. I turn to Pro Search when I need a more robust answer than a simple search can provide, but don’t want to wait 20 minutes for a long, deep-dive report.
💵 Paid:
Google Deep Research (with Gemini 2.5 Pro Experimental) [NEW]: Paying Gemini Advanced users get to experience the current best “Deep Research” product on the market, powered by Google’s most powerful model.
📈 6. Productivity
🎁Free:
Napkin [NEW] can automatically turn walls of text into catchy visuals like diagrams, infographics, etc. For design-challenged people like myself, it’s a superb way to illustrate concepts with minimal effort.
💵 Paid:
Genspark Super Agent [NEW]: A robust agent that can orchestrate many tools to complete complex, multi-step tasks. You get 200 free monthly credits to try it, but you’ll likely need the paid plan for bigger tasks.
People often use the term “photorealistic” here, but I take issue with it. Photorealism is used specifically to describe something that’s not a photograph (e.g. a painting) but is made to look like one. So I use “photographic” to refer to AI images that faithfully mimic the look of an actual photo.
I don’t use it much myself, but it’s an easy recommendation based on my tests and everything I’ve seen from other users.
Even though I crowned a winner in my Sunday Showdown #6.