Mistral Releases Open-Weight Safety Classifier Shieldstral 1.0 3B
A lightweight model that turns policies into text-based questions, letting operators change content moderation standards without retraining
10
Mistral AI
A French model developer. It positions itself around "sovereign AI" that processes data within Europe. Its flagship product is Shieldstral 1.0 3B, an open-weight safety classifier released in August 2026. That same month, it formed a compute coalition under the banner of European AI sovereignty, opening it up even to external models. It also introduced a feature that processes data on European servers, though agentic features were excluded from that scope. In the European AI sovereignty debate reported in August 2026, some critics pointed out that this amounted to little more than slapping a European zip code on American-made models.
Current rank (1M)
32
- Change
- Prev.
- 28
Rank over the last 30 days
2026.08.18 – 2026.09.16 · High 25 · Low 32 · Now 32
Mistral AI · Related stories
Anthropic Sets Up In-House AI Chip Design Team
Anthropic has officially confirmed it is building an internal custom semiconductor design team to power its own models. Job postings for senior silicon engineers and technical program managers are already live.
20
That's the last story.