# Sagar Vaze

Markdown copy of https://realalpha.app/people/sagar-vaze/ on Real Alpha, the best minutes from long interviews with the people building AI. Cite and link the page, not this copy.

Real Alpha has cut 6 moments of Sagar Vaze (Research Scientist, Mistral) talking, from an interview on MTS published Oct 29, 2026.

- Role: Research Scientist, Mistral

## What Sagar Vaze said

- [Closed-model refusals can leave cyber defenders without recourse](https://realalpha.app/s/closed-model-refusals-can-leave-cyber-defenders-without/)
  MTS · Oct 29, 2026 · 3 min
  > “these defensive requests in practice look quite similar to the offensive ones.” — Sagar Vaze · [16:03](https://realalpha.app/v/gJZLJSIEbtQ/?t=963)
- [Mistral trained its trillion-parameter model on its own compute](https://realalpha.app/s/mistral-trained-its-trillion-parameter-model-on-its-own/)
  MTS · Oct 29, 2026 · 3 min
  > “It's a trillion parameter model trained on 1E25 flops, which is a measure of the amount of compute that went into it.” — Sagar Vaze · [10:09](https://realalpha.app/v/gJZLJSIEbtQ/?t=609)
- [Fine-tuning can beat prompting on narrow enterprise tasks](https://realalpha.app/s/fine-tuning-can-beat-prompting-on-narrow-enterprise-tasks/)
  MTS · Oct 29, 2026 · 2 min
  > “doing a small, what we call lower fine tuning, or a small fine tuning of the model on the data which we'd collected for this exact task was way, way better.” — Sagar Vaze · [4:39](https://realalpha.app/v/gJZLJSIEbtQ/?t=279)
- [Compute is Mistral's main gap to the leading AI labs](https://realalpha.app/s/compute-is-mistrals-main-gap-to-the-leading-ai-labs/)
  MTS · Oct 29, 2026 · 2 min
  > “we've convinced ourselves that we can do more with less.” — Sagar Vaze · [7:35](https://realalpha.app/v/gJZLJSIEbtQ/?t=455)
- [Open weights let firms turn proprietary expertise into a moat](https://realalpha.app/s/open-weights-let-firms-turn-proprietary-expertise-into-a/)
  MTS · Oct 29, 2026 · 2 min
  > “imagine you're an engineering company in Germany rather than being worried that the next generation of closed-weights models is going to come and wash over your competitive advantage.” — Sagar Vaze · [2:40](https://realalpha.app/v/gJZLJSIEbtQ/?t=160)
- [Mistral trained multimodal capabilities from the start](https://realalpha.app/s/mistral-trained-multimodal-capabilities-from-the-start/)
  MTS · Oct 29, 2026 · 2 min
  > “rather than training a model on text and then fine-tuning it on images and text or videos and text from the very beginning it's got images baked into it” — Sagar Vaze · [18:07](https://realalpha.app/v/gJZLJSIEbtQ/?t=1087)

## The episode

- [Why Closed AI Providers Block Cyber Defense | Sagar Vaze](https://realalpha.app/v/gJZLJSIEbtQ/) · MTS · Oct 29, 2026