What changed
Mistral released Small 3.1 under Apache 2.0 with image understanding, a 128k context window, function calling, and local deployment guidance. It retained the 24B size of January's Small 3.
Why it matters
A relatively compact, permissively licensed model could cover document images, local assistants, and tool-using applications without a separate vision stack.
What to do
Evaluate it on screenshots, scanned documents, charts, and long mixed-media inputs, not only text benchmarks; test both full and quantized checkpoints.
Caveat
Local hardware feasibility and speed depend on precision and memory. Comparative benchmark claims were Mistral-run.