Mistral Small 4, released in March 2026 as a 119B-parameter MoE model unifying instruct, reasoning, and multimodal capabilities under Apache 2.0, set the current baseline with its 256k context and efficient 6B active parameters. Trader focus on a 4.1+ successor stems from Mistral’s pattern of iterative Small updates roughly every 3–6 months, now overshadowed by the October 6 public preview of Large 4 (“Le Chonk”), a 1T-parameter frontier model slated for open weights by month-end. Competitive pressure from labs like Qwen and frequent releases elsewhere, plus Mistral’s €3B funding round scaling European compute, heighten uncertainty around timing for the next Small iteration. Key near-term catalysts include Large 4’s full rollout and any developer conference or earnings signals on roadmap adjustments.