DeepSeek-V3.2-Speciale is a high-compute version of DeepSeek-V3.2, designed for maximum reasoning and agentic performance. It builds on DeepSeek Sparse Attention (DSA) for efficient long-context handling and uses scaled reinforcement learning post-training to surpass the base model. Evaluations show that Speciale outperforms GPT-5 on complex reasoning tasks and matches Gemini-3.0-Pro in proficiency, while maintaining strong coding and tool-use reliability. Like V3.2, it uses a large-scale agentic task synthesis pipeline to enhance compliance and generalization in interactive environments.
Not enough throughput data
Not enough TTFT data