Generative AI has come a long way in the past year, going from a handful of AI image generators, ChatGPT and the first signs of AI video to an ecosystem of creativity. 

Every big tech company has fully embraced the world of generative AI and chatbots like ChatGPT, Claude and Gemini can hold their own in everything from generating computer games to writing impressive poetry.

We've seen the emergence of advanced, natural-sounding generative music complete with vocals from companies like Suno and Udio and AI video is now so real it's becoming indistinguishable from reality.

A handful of companies in this space stand out from the crowd, often offering unexpectedly impressive features, a strong community system, or advanced capabilities not available anywhere else and these are our favorites.

In just under a year Claude has gone from being something of an underdog compared to ChatGPT to one of the most powerful and feature-rich AI productivity platforms on the market. 

With Sonnet 3.5 it is capable of advanced reasoning, complex coding, an ability to understand even the most obscure images, and an improved understanding of humor. 

What makes Claude stand out for me as the best AI chatbot of the year is the constant upgrades and new features from Anthropic, among them Artifacts — the best AI feature of the year from any company. This allows you to interact with a game, app, code or graphic generated by Claude without leaving the interface and even share the creation with others.

Here at Tom’s Guide our expert editors are committed to bringing you the best news, reviews and guides to help you stay informed and ahead of the curve!

On top of Projects and an improved user interface, Anthropic has also added a Projects feature that lets you load a range of files and Artifacts linked to a single chat.

Artificial Intelligence image generators came into their own in 2024, reaching levels of realism not previously possible and even developing the ability to accurately render text on an image. There are more AI image generators than any other type of generative content product and for me, Leonardo stands out as the most feature-rich and original product in a crowded market.

As well as Stable Diffusion-based models to generate complex and photorealistic images, upscaling and even texture generation, Leonardo has its own model in Phoenix. What this new generator can do is follow a prompt unlike any I’ve tried before it. You can be explicit in what you want, even with complex queries, and it will accurately render the result.

Other features include video generation from any image, real-time generation and even the ability to create an AI image from a rough sketch rather than a text prompt.

Competition in the AI video generation space grew rapidly in the past 12 months, going from just Runway out on its own as the only commercial synthetic AI video tool, to an increasingly crowded market that includes Pika Labs, Luma Labs Dream Machine, Haiper, LTX Studio and, at least for filmmakers if not the general public, Sora from OpenAI.

The reason Runway is my AI video generator of the year, despite a crowded market is the fact it has continued to innovate, including a new model capable of impressive degrees of realism in both motion and video quality. Gen-3 isn’t fully live yet, available as an Alpha release and in text-to-video only, but despite that it is the best publicly available video model on the market.

On top of its impressive video generation, Runway’s lip-sync is one of the most impressive I’ve used. The site comes with asset catalogs and team-based interfaces and you have a range of other image, 3D and audio features available as part of the same subscription.

There are a number of AI music generators and this may be one of the most controversial categories due to concerns, and legal action, from the music industry. There are generators that use only licensed and tracked music and offer original features but Udio is by far the best sound quality, musical realism and vocal generation of any currently publicly available.

It was very close between Udio and Suno, but in the end the ability to set specific generation points within a song as you extend it, customization of lyrics and styles and a growing community won out for me. Udio songs also capture a sense of emotion others struggle with.

Udio is also more controllable than other models. You can use tags to set certain parts of the song as speaking, create cross-genre tracks and build out a song from your own sounds.

The Meta Ray-Ban smart glasses have been elevated in just a few months from a pair of camera glasses with some voice assistant stuff built in, to a full-blown AI device that genuinely brings big benefits to your day-to-day.

Yes, that $299 asking price may be a little steep for a pair of specs, but the Meta tech inside and the regular software updates make these more than worthwhile. It all starts with the 12MP camera on the front that captures some impressive photo and video content in an Instagram-friendly 9:16 aspect ratio. Alongside this, impressive built-in speakers and microphones make this great for taking calls and ambient music on-the-go.

But the real beauty comes in its AI, as the smarts unlocked in this form factor are massive. For example, during my time in Costa Rica, I was able to look at a tourist landmark and ask for details about it — which the Meta Ray-Bans were happy to oblige with details. 

Or at a restaurant where I was instantly bewildered by a long Spanish menu, I received an audio translation of what’s on the piece of paper. Put simply, this is the best piece of AI hardware you can buy right now.

Imagine Google Lens, but powered by AI and considerably faster and more convenient to use. That’s a basic explanation of Circle to Search on Android. With a few quick gestures, which includes drawing the titular circle, you can ask Google AI what it is you’re looking at. All without having to switch to a different app, type or take any screenshots. 

The AI is pretty subtle, and you might not have realized Circle to Search was actually using AI. But it is there, identifying objects and sounds, translating text into something you can read, and answering questions you might have about what’s on your screen. The best part is that Circle to Search is so easy to use you can easily activate it without even realizing.