It feels like every week there’s a new AI model that’s supposed to be the next big thing, right? The space is getting incredibly crowded, and honestly, it’s hard to keep up. We have the giants like OpenAI and Anthropic who seem to be in a constant arms race, releasing bigger and better models every few months.
So when a new lab pops up on the scene, you have to wonder: can they actually compete?
Well, a company called Thinking Machines Lab just threw its hat in the ring, and I have to say, their first move is a bold one. They’ve just released a new open-source model named "Inkling," and it’s not just another text-based chatbot. This one is a little different, and it’s got some serious muscle behind it.
Let's get into what makes this announcement so interesting.
So, What Exactly is This 'Inkling' Model?
Alright, let's break down the specs. Inkling is a massive 975-billion-parameter model.
Now, I know "parameters" can sound like a bunch of technical jargon. The easiest way to think about it is like the number of knobs and dials the AI can tune while it's learning. More parameters generally mean the model has a greater capacity to learn complex patterns and nuances from the data it's trained on.
For a little perspective, some of the most powerful models out there are in the trillion-parameter club. So, at 975 billion, Inkling is playing in the big leagues right out of the gate. This isn't some small-scale experiment; Thinking Machines Lab is clearly aiming to make a statement.
But the size isn't even the most fascinating part.
It's Not Just About Text Anymore
Here’s the thing that really caught my eye: Inkling was trained to understand video and audio.
Most of the AI models we interact with daily—think ChatGPT or Claude—are primarily text-based. You type a question, they type an answer. They’re brilliant at understanding and generating language, but their world is made of words. They don’t really experience the world the way we do, through sight and sound.
Inkling is built differently. By training it on video and audio, Thinking Machines is trying to give its AI a more holistic understanding of information.
Imagine this for a second:
- You could show it a video of a mechanic fixing a car engine and ask it to generate a step-by-step repair guide.
- You could have it listen to a recording of a business meeting and have it instantly pull out the key decisions and action items.
- It could even watch a silent film and write a plausible script for what the characters might be saying.
This is what we call "multimodal" AI, and it’s a huge step toward creating AI that can interact with the world in a much more human-like way. It’s about moving beyond the text box and into the rich, messy, and dynamic world of sensory data.
The Open-Source Gambit
Here’s another huge piece of the puzzle: Thinking Machines Lab released Inkling as an open-source model.
This is a major strategic move, and it puts them in direct contrast with some of their biggest competitors. Companies like OpenAI and Anthropic keep their most powerful models under lock and key. You can use them through an API, but you can't look under the hood, modify them, or run them on your own servers. They’re a black box.
By making Inkling open source, Thinking Machines is inviting the entire global community of developers and researchers to play with their new toy. They can dissect it, build on it, find its flaws, and discover new ways to use it.
This approach does a couple of things:
- It accelerates innovation. A thousand smart people experimenting with your model will find new applications way faster than a small internal team ever could.
- It builds trust and community. Developers tend to gravitate toward tools they can freely access and understand. It’s a great way to build a loyal following.
- It’s a direct challenge. It’s a way of saying, "We’re so confident in our tech that we’re not afraid to share it. Let's see what you can build."
Can They Really Take on the Giants?
So, this is the big question, isn't it? Can a newcomer like Thinking Machines Lab truly establish itself alongside the titans of the industry?
It's definitely an uphill battle. OpenAI and Anthropic have enormous funding, massive teams of top-tier researchers, and a significant head start. They’ve built powerful brands that people know and trust.
But Thinking Machines has a few things going for it with this Inkling launch. The focus on video and audio is smart—it carves out a specific niche where they can lead. And the decision to go open source is a powerful differentiator that will attract a lot of talent and attention.
It reminds me a bit of the early days of Linux versus Windows. One was a closed, corporate product, and the other was a community-driven, open project. We all know how that turned out—both found massive success in different ways.
I don’t think Thinking Machines is going to put OpenAI out of business tomorrow. But they don't have to. By releasing a powerful, open, and unique model like Inkling, they've officially put themselves on the map. They've started a conversation and given the AI community a fascinating new tool to work with.
For us, it’s just exciting. More competition means faster innovation, more choices, and more pressure on the big players to keep pushing the boundaries. I’ll be watching Thinking Machines Lab very closely to see what they do next. This is one underdog that just might have a serious bite.
