Mistral AI released Small 4 in March 2026 as an open-weight MoE model with 119B total parameters and roughly 6B active, unifying instruct, reasoning, multimodal, and agentic coding capabilities under Apache 2.0 with a 256k context window. Trader attention now centers on the absence of any confirmed Small 4.1 or higher update amid Mistral’s October 2026 push of Large 4 into public preview, with weights scheduled for October 27. Competitive pressure from efficient open models by Qwen and other labs, plus Mistral’s emphasis on larger frontier releases and specialist variants like OCR 4.1, shapes expectations for the next Small iteration. Key near-term signals include developer conference timelines, benchmark releases, and any hints from Mistral’s infrastructure partners on release cadence.