Mistral Small 4, released March 16, 2026, unified instruct, reasoning, multimodal, and agentic coding capabilities in a 119B-parameter MoE architecture with 6B active parameters, 256k context, and Apache 2.0 weights, delivering 40% lower latency than Small 3. As of mid-October 2026, no official announcements or leaks have surfaced for a 4.1+ successor despite Mistral’s active cadence, including Large 4’s October preview and Magistral Small 1.2 updates. Traders monitor Mistral’s release patterns, competitive pressure from models like Qwen and GLM variants, and signals around reasoning_effort tuning or multimodal refinements that could accelerate the next Small iteration. Upcoming developer events or API changelog entries remain key near-term catalysts.