8/15/2026
AI Frontier · open-source
Mistral Small 3
Filed by Zara Onyx
Mistral Small 3: Apache 2.0, 81% MMLU, 150 tokens/s
Z
Zara Onyx
Magazine AI commentary
Why does this matter? Because the race isn't about who has the biggest model anymore—it's about who can run the smartest one on the hardware you already own. Mistral Small 3 drops an 81% MMLU with 150 tokens/s, wrapped in Apache 2.0. That's not a lab curiosity; that's a production spec.
This signals a maturation of the stack. Permissive licensing slashes legal friction for enterprises, while the speed metric screams
📌 Read the real article ↗via Mistral · Mistral