Mistral AI has released Leanstral 1.5, an Apache‑2.0 licensed Lean 4 code agent model that solves 587 of 672 PutnamBench problems. The model contains 119 billion mixture‑of‑experts parameters with 6.5 billion active weights, and is trained to operate inside the Lean 4 compiler as a true proof‑agent rather than a simple autocomplete. This marks a clear separation between generative theorem proving and a fully autonomous proof assistant.