Meta quietly building a new AI model — here's what to expect

The logo for Meta's Llama 3 AI
(Image credit: Meta)

The arms-race between technology companies regarding AI models has heated up significantly in the last two weeks. OpenAI revealed GPT-4o immediately before Google announced a number of Gemini improvements. That was followed by Microsoft announcing a bunch of Copilot+ PCs along with AI improvements of its own.

But while all that was going on, Meta was taking care of AI business of its own. The company quietly released a research paper on its efforts in the multi-modal AI space. Spotted by Venture Beat, the paper shows that Meta is working on a multi-modal large language model called Chameleon.

That's not to be confused with the generative AI model CM3leon (pronounced Chameleon) that Meta AI revealed last summer. Meta AI did say in its blog post that the CM3leon model would lead to improvements on future LLMs. 

The research paper claims that Chameleon is state-of-the-art and beats or competes evenly with other models like Gemini, GPT-4 and Meta’s own Llama-2. 

Similar to Google’s Gemini, Chameleon is built on an “early-fusion token-based mixed-modal” architecture. This means that model was built to learn from the beginning from a combination of images, code, text and other inputs, and it uses that content to create sequences.

The other way to build a multi-modal architecture is to sew together a number of models that are trained on a single modality. This is called “late fusion.” In essence, the AI system takes individual models and fuses them together to make inferences. Late fusion apparently works well but can be limiting in the ability for the AI to integrate information.

In the paper, the authors say that Chameleon is most similar to Google’s Gemini which was built in a similar fashion. Unlike Gemini, though, the researchers say that Chameleon is an end-to-end model. 

If the paper’s claims and testing are true and can be replicated, it does appear that the Chameleon model matches or exceeds many of the AI models that are available today. 

An interesting wrinkle to the claims made about Chameleon, is that Mark Zuckerberg and Meta have been pushing open source as the future for much of this year. The already existing Llama 3 is open source and expected to receive big update in July. Meta also just opened up the Quest headset operating system to hardware manufacturers. 

The paper does not indicate if or when Meta might release this new model. Publicly, Meta has been hard at work on the latest iteration of the Llama assistant. Llama 3 just went live on Facebook, Instagram and WhatsApp in late April.

More from Tom's Guide

Scott Younker
West Coast Reporter

Scott Younker is the West Coast Reporter at Tom’s Guide. He covers all the lastest tech news. He’s been involved in tech since 2011 at various outlets and is on an ongoing hunt to build the easiest to use home media system. When not writing about the latest devices, you are more than welcome to discuss board games or disc golf with him. 

Read more
Meta AI logo on a phone
Meta AI could take on ChatGPT and Gemini with standalone app launching within months
Meta AI logo on a phone
Meta set to release a direct competitor to ChatGPT — here's what you need to know
Copilot, Gemini, Claude
I test AI chatbots for a living and these are the best ChatGPT alternatives
OpenAI logo on a phone screen
OpenAI's next generation model could be announced today — here's what we know
Facebook co-founder, Chairman and CEO Mark Zuckerberg arrives to testify before the House Energy and Commerce Committee in the Rayburn House Office Building on Capitol Hill April 11, 2018 in Washington, DC. This is the second day of testimony before Congress by Zuckerberg, 33, after it was reported that 87 million Facebook users had their personal information harvested by Cambridge Analytica, a British political consulting firm linked to the Trump campaign.
Meta acknowledges 'critical risk' AI systems that are too dangerous to develop — here's what that means
Google Gemini
Google Gemini — everything you need to know
Latest in AI
Google Gemini logo
You can now use Google Gemini without an account — here's how to get started
Slack
ChatGPT could soon work with Slack and Google Drive to make office life a bit easier
Gemini and ChatGPT logos on a phone
Forget ChatGPT Canvas — I just tried Gemini Canvas and I'm floored by the difference
Gemini screenshot image
Gemini just became the ultimate collaborator — everything you need to know about this huge new upgrade
Glok vs. Claude logo for AI madness
I tested Grok vs. Claude with 5 prompts to crown a winner
Google Assistant logo on a smartphone screen
Google Assistant is losing features to make way for Gemini — here's what's just been axed
Latest in News
A phone with the Plex logo in front of an out-of-focus background of movie posters
Yikes! Plex is getting a price hike and this key feature is going behind a pay wall
back of Iris Pixel 9a
Google Pixel 9a pre-orders delayed due to 'component quality issue' — here's when you can get one
An open lock depicting a data breach
Half a million teachers hit in major data breach with SSNs, financial data and more exposed — what to do now
Sony A95K QD-OLED TV in front of windows in a living room
This new TV breakthrough looks like a game-changer for OLED TVs
Apple iPhone 16 & 16 Plus hands-on.
Forget USB-C — a truly portless iPhone just got the all-clear from the EU
Samsung Galaxy Z Flip 6 features on outer cover display
Samsung Galaxy Z Flip FE may arrive 'months' after the Z Flip 7 — here's why