Mistral Small 4 launched in mid-March 2026 as an efficient 119B-parameter MoE model with 256k context, unifying instruct, reasoning, multimodal, and agentic coding capabilities previously split across Magistral, Pixtral, and Devstral lines. Since then, the company has prioritized larger releases, including Medium 3.5 in April and the frontier Mistral Large 4 preview in early October, with open weights expected by month-end. This pattern of quarterly-to-bimonthly major drops, combined with ongoing optimization for cost and latency against competitors like Qwen and Claude Haiku variants, shapes trader views on when a 4.1+ iteration might appear. Key watchpoints include post-Large 4 refinements, developer feedback on Small 4 benchmarks, and any signals around Mistral’s next efficiency-focused update.