Speciale, claiming GPT-5 and Gemini-3 Pro-level reasoning. New models post top results on math and coding benchmarks.

DeepSeek introduces V3.2 and V3.2 Speciale models with advanced reasoning abilities, aiming to match GPT 5-level performance and challenge Google’s Gemini 3 Pro across major benchmarks.

is a senior reporter who has covered AI, robotics, and more for eight years at The Verge. DeepMind has created an AI system named AlphaCode that it says “writes computer programs at a competitive ...