Generative AI has come a long way in a very short space of time, from startups dabbling in image generation to an entire industry of AI video, music, chat, voice and image tools used by everyone from filmmakers to influencers.
As part of my work covering AI for Tom's Guide, I try out, test, and push the limits of new and updated AI tools. Some stick around and I use them every day, some surprise me and I can't help but talk about them, and some just disappear.
Creating a specific list of the best AI tools overall was a nearly impossible task, as each product category has a variety of options offering different tools and styles, but these are the five tools I use every day as part of my job and for fun.
The only category where this was an obvious choice was synthetic voice: there are alternatives, but none come close to the breadth and quality of ElevenLabs.
The tools that end up on this list have to be ones I use every day, or at least most of the time on the weekends because sometimes I just want to get away from technology.
It has to be easy to learn, relatively easy to use for a first-time user, and of the highest quality possible, taking into account other factors. So if it's great quality but unusable, it won't be adopted.
The final point is pricing – there needs to be a free tier with either a reasonable number of credits at launch, or credits that reset. This is why Leonardo beats Midjourney on my list.
| Header Cell – Column 0 | product | Free Plan Credits | Cheapest paid plan |
|---|---|---|---|
| image | Leonardo | 150 per day | $12 |
| video | LumaLab Dream Machine | 30/month | $29 |
| music | audio | 10 times a day + 100 times a month | $ 10 |
| chat | Claude 3 | N/A | $20 |
| voice | Eleven Lab | 10,000 characters/month | 5 dollars |
The best AI image generator
Leonardo, like many others in its category, started life as a very well-made wrapper for various Stable Diffusion models, but has since started to roll out its own models, including the impressive Phoenix, which comes with custom styles and tweaked versions of the models, as well as other AI image tools, all of which make it a standout in its field.
With the fine-tuned PhotoReal model, the ability to generate photorealistic images is almost up to Midjourney's standard, and you can create a variety of styles using the Elements feature.
These elements are fine-tuned models that can be applied before generation, controlling the resulting image to have a particular look, such as sketched or sculpted, or you can set a style such as cinematic, food, long exposure, etc.
What really makes Leonardo stand out to me is the combination of an easy to use UI and incredible control: you can add reference images and configure how the AI uses them, control the size and layout, even add transparent backgrounds.
Most of these features are available on other platforms, but Leonardo has all of these features plus much more, including image upscaling, live image generation, and one of its most creative tools: the ability to draw a sketch and have the AI transform it into a perfect image.
The best AI video generator
Dream Machine came out of nowhere, coming from a company that had previously focused on generative 3D content. The Luma Labs Genie model was a major milestone in converting text to 3D models, and it looks like the company has leveraged some of that knowledge to create text-to-image and text-to-video models.
Demand for Dream Machine was so high when it first launched that the company had to quickly implement a five-generations-per-day limit for free users, and it's also seeking more computing power to run models on social media.
Each video generated is about 5 seconds long and has impressive following prompts: Giving explanatory ideas will improve the model to get the best results.
Shortly after release, they also added the ability to extend clips by up to five seconds, but in my experience this is a mixed bag: when it works, it's easy, but if you don't enter the prompts exactly right, it makes some strange changes to your original video.
Videos created with the Luma Labs Dream Machine are just as realistic as the ones we saw in the Sora examples, and the level of motion control is impressive — unlike Sora, I was able to see this in videos I created myself. It's easy to use, enhances your own prompts, and works well with traditional filmmaking cues like dollies.
Luma Labs Dream Machine's free plan allows you to create 30 videos per month, which you'll quickly run out of if you want to do more than just play around. Paid plans start at $30 per month and allow you to create 120 more in addition to the free 30. They also remove watermarks, enable commercial use, and give you higher priority in the queue.
The best AI music generator
Udio was launched earlier this year and wowed the AI world with its incredibly natural vocals – previous models sounded a bit autotuned in the voice parts, though music was generally better.
What I like most about Udio is the amount of control you have over the song that is generated: although you can only create the first 30 seconds of a clip, you can extend this by adding intros, new sections before and after, and outros.
It can use custom lyrics, generate lyrics, or create instrumental tracks, and it works across a wide range of genres, including stand-up comedy.
In addition to the ability to generate songs, Udio has a growing community that gives you easy access to a library of tracks made by other users, each of which you can view and in some cases reuse prompts in your own songs.
One of Udio's newest features is the ability to extend your own sounds – I tried this with a range of noises, including Microwave, and the company says it's working on the ability to embed sounds into tracks.
The free plan gives you 10 credits per day, plus an additional 100 credits once a month, but doesn't allow you to use your songs commercially.
Best AI Chatbots
Claude is the most human chatbot I've ever interacted with. Not only is it a great alternative to ChatGPT, but overall it's better than ChatGPT at the moment. It's good at reasoning and persuasion, and it's not that lazy. It will create a complete app or write an entire story. Sonnet 3.5 is even more entertaining and visually improved.
Claude 3 has one of the largest context windows of any AI chatbot, with around 200,000 by default and scaling up to 1 million for certain use cases. This is especially useful now that Claude 3 has vision capabilities and can easily analyze images, photos, graphs, and with Sonnet, it can even understand handwritten text.
The free version of Claude 3 comes with Sonnet, which has recently been significantly upgraded and outperforms the larger Opus models for most tasks. It has speed limits, but above those limits it falls back to Haiku, which is faster but has some inference limitations.
Claude 3 doesn't have image generation capabilities, but it can generate vector graphics if you try hard enough. It's particularly good at providing prompts that you can paste into image generators like Midjourney. It also has better coding than the other models.
The latest update introduces Artifacts, a new sidebar feature that lets you view the code, text, web pages, and images you've created.
The best AI voice generator
ElevenLabs is the best, easiest to use AI sound platform to date, evolving from its original voice-centric focus to sound effects, voice-overs, and soon music.
There are many very good synthetic voice platforms available, but they tend to be aimed at the enterprise market, difficult to use, and have slow or low accuracy when it comes to replicating voices.
With just a three-minute audio recording, you can create an exact clone of your voice and use it to create your own podcast or read a report without actually saying a word yourself.
Features like Dubbing Studio allow you to translate into different languages, and Voice Over Studio lets you add different voices to your videos, allowing you to add speakers and sound effects to your silent movies.
I think the best thing about ElevenLabs is their vast library of voices, all licensed and free to use.
This includes premium voices of actors and some celebrities, all of whom are paid when you use their voices (the payment is made by ElevenLabs, not directly), such as Christy Carlson Romano, the TV actress and voice of Disney's Kim Possible (though you need special permission to use her voice).
The free plan is incredibly comprehensive, giving you more than enough characters to really play with across all the available features, and paid plans start at just $5.
Want to learn more about using AI for creative work? The best AI image generator or list The best AI video generator.
Back to MacBook Air
