# Efficient, task-specific models can compete on less funding

Markdown copy of https://realalpha.app/s/efficient-task-specific-models-can-compete-on-less-funding/ on Real Alpha, the best minutes from long interviews with the people building AI. Cite and link the page, not this copy.

- Speaker: [Marjorie Janiewicz](https://realalpha.app/people/marjorie-janiewicz/) · Marjorie Janiewicz is chief revenue officer at Mistral.
- Host: [Theo Jaffee](https://realalpha.app/people/theo-jaffee/) · Co-founder & host, MTS · One of the first members of a16z's new media team; earlier hosted the Emergent Ventures-backed Theo Jaffee Podcast.
- Host: [Sophia Dew](https://realalpha.app/people/sophia-dew/) · Host, MTS
- Show: MTS
- Episode: [You Don't Need Billions to Build Frontier AI | Marjorie Janiewicz](https://realalpha.app/v/4X3dZmBPIRY/), Oct 30, 2026
- Moment: 11:40 to 13:41 of the episode (2 min), playing on https://realalpha.app/s/efficient-task-specific-models-can-compete-on-less-funding/; the episode from there: https://realalpha.app/v/4X3dZmBPIRY/?t=700

## Quote

> “And it does prove the point that you **don't need billions of dollars** to build a frontier model that's as strong as closed source models.”
> — [Marjorie Janiewicz](https://realalpha.app/people/marjorie-janiewicz/) at [12:11](https://realalpha.app/v/4X3dZmBPIRY/?t=731) · MTS

## Takeaways

1. Mistral had raised roughly $6 billion, she says.
2. Open weights and model design help reduce training costs.
3. Task-specific models can be smaller and more efficient.

## About this segment

Marjorie Janiewicz credits open weights and design choices for building competitive models with fewer resources. Smaller models adapted to specific tasks can also lower customer costs.

From “You Don't Need Billions to Build Frontier AI | Marjorie Janiewicz”, Oct 30, 2026. [Watch the full episode](https://realalpha.app/v/4X3dZmBPIRY/)

## Transcript

Transcribed by machine (Whisper) from the episode’s audio, so a word may be off here and there.

[11:44](https://realalpha.app/v/4X3dZmBPIRY/?t=704) about how you're able to build this with a fraction of the resources. I mean, famously, a lot of other competing labs were able to raise hundreds of billions of dollars. I think up to today, you guys have raised roughly $6 billion. You're also able to compete with fewer resources than maybe some of these other frontier labs. Tell us about how that's possible. It's possible because of the open weight approach. It's possible because of the design choices that we make. And it does prove the point that you don't need billions of dollars to build a frontier model that's as strong as closed source models. And that value that we, I mean, unlock for our own model training, we bring that back to our customers. Customers are starting to worry about their bills, you know, the token price and how much investment is going into their AI.

[12:44](https://realalpha.app/v/4X3dZmBPIRY/?t=764) When Mistral is able to really build those models at a fraction of the cost, It does mean that it provides customers the ability to also leverage smaller models that are more efficient. And in turn, they are going to see some of that value back in the cost efficiency by leveraging Mistral. So it is possible when you build efficient, smaller models that are task specific. As an example, we release a time series models for financial services like best of breed, best model, state of the art.

[13:16](https://realalpha.app/v/4X3dZmBPIRY/?t=796) It's very small. It's very agile. It outperforms all models in time series context when it's adapted to our customers' data. It's very powerful. And it means that a lot more is going to come with me strong. A lot of great models that are really competitive and that are truly giving an alternative to customers. And it's open. And it's all open. I mean, yeah, I mean, we talked about this a lot.