ByteDance has announced the release of Doubao Large Model 2.0, featuring three general-purpose Agent models—Pro, Lite, and Mini—and a specialized Code model. Doubao 2.0 Pro is tailored for deep inference and long-chain task execution, rivaling GPT 5.2 and Gemini 3 Pro. The 2.0 Lite model offers a balance between performance and cost, outperforming the previous Doubao 1.8. The 2.0 Mini model is optimized for low-latency, high-concurrency, and cost-sensitive applications. Additionally, the Doubao-Seed-2.0-Code model is designed for programming tasks, delivering enhanced results when paired with TRAE.