Thinking Machines — a lab founded by OpenAI's former CTO — released Inkling Small, a reduced version of the Inkling launched the week before.
It's an omnimodal model: it understands text, audio, image, and video in the same model. It's about a quarter the size of the original, which still leaves it at 276 billion total parameters, with 12 billion active per use.
Why this matters
- Audio is the modality where open models usually underperform — and it's exactly where this one stands out.
- In some tests, it outperforms the full model using less compute.
- It's one point away from the full Inkling in independent evaluation, at a much lower cost.
1.Be honest about where it loses
2 minThe source doesn't sell the model as an overall champion, and it's worth repeating the caveat:
In terms of intelligence, the Inkling Small isn't the best, but I like its multimodal capabilities.
Spec sheet
- License
- Aberto, no Hugging Face
- Parameters
- 276 B totais / 12 B ativos
- Formats
- Text, audio, image, and video
Continue in the full microcourse
You've read the opening of 3 classes
The microcourse covers the complete step-by-step, the selection criteria, where the tool fails, who it's really for — and, in the Expert version, the official address to start today.
- When this model is the right choice2 min
- The cost of running2 min
How we verified
We track releases straight from primary sources, transcribe what's demonstrated, check every name and number against the manufacturer's official documentation, and rewrite it in Portuguese — with what the tool no do it together, which is the part the ad leaves out.
