Ministral
Mistral's 3B and 8B edge models for on-device and low-latency use
About
Ministral 3B and 8B arrived October 16, 2024, on the first anniversary of Mistral 7B, as Mistral's first models designed specifically for the edge. Both carry a 128K context window — the 8B uses an interleaved sliding-window attention pattern for speed and memory efficiency — and target on-device translation, offline assistants, local analytics, and robotics. Mistral reported them beating Gemma 2 2B, Llama 3.2 3B, and even Mistral 7B on knowledge, commonsense, and function-calling benchmarks. The licensing is the wrinkle: weights are downloadable but commercial use requires a Mistral commercial license, so they are open-weight rather than open-source. Qualcomm partnered to optimize both for Snapdragon platforms.
Specs
Context window
Pricing
Free quota
API access
Weights
Released
Benchmarks
Team-reported: outperformed Gemma 2 2B, Llama 3.2 3B, and Mistral 7B on knowledge, commonsense, and function-calling (team-reported)
Provider-reported unless a source says otherwise. Benchmarks are marketing-adjacent by nature — treat them as directional.
Use cases
Trust score — 7/10
Scored 7/10 (verified tier, metrics-v1). Evidence: Released 2024-10-16 per Mistral and press; Weights downloadable but commercial use requires Mistral commercial license — caveat captured; 3B and 8B variants, 128K context, edge-focused (team specs); Benchmark wins are team-reported; Qualcomm Snapdragon partnership verified.