Mozilla has added Mistral Small 4 as a selectable model in the beta Smart Window in Firefox, its first named partnership with Mistral AI, while Mistral continues to ship open-weight models, most recently the Shieldstral 1.0 3B moderation model in August 2026.
Mistral AI's recent record consists of two product releases: a moderation model and a model partnership with Mozilla. On August 4, 2026, the company released Shieldstral 1.0 3B, a 3-billion-parameter model that judges whether text or images meet safety standards specified in natural language. It supports 12 languages, runs on a single GPU with 16GB VRAM, and carries an Apache License 2.0.
In September 2026, Mozilla entered a partnership with Mistral AI. Under that agreement, Mistral Small 4 can be selected as a model in Firefox's beta Smart Window, a test window alongside the normal and private windows that groups tabs, hosts chat, and answers questions. Mozilla says it chose the open-weight, Apache-licensed European model for multilingual performance, and added French as a supported language.
Across both stories, the throughline is language coverage and open licensing. Shieldstral 1.0 3B covers 12 languages under Apache 2.0. Mistral Small 4 was picked for Firefox on multilingual performance and is described as open-weight and Apache-licensed. Mozilla cites further European expansion in 2026, but the two stories do not otherwise state what other integrations, releases, or model families Mistral has in the works.
Synthesized by Yomimono from the cited Yomimono stories below, each itself
sourced, then editorially reviewed. Every
fact links the story it came from.
Sep 17
Mozilla has entered a partnership with France-based AI developer Mistral AI. Under the agreement, Mistral Small 4 can be selected as a model in Firefox's beta Smart Window, a test window alongside the normal and private windows that groups tabs, hosts chat, and answers questions. Mozilla cites multilingual performance and has added French as a supported language, with further European expansion planned in 2026.
Aug 10
Mistral AI released Shieldstral 1.0 3B on August 4, 2026, a 3-billion-parameter AI model that judges whether text or images meet safety standards specified in natural language. It supports 12 languages, runs on a single GPU with 16GB VRAM, and is available under Apache License 2.0.