DeepSeek
Deepseek V4 Flash
DeepSeek-V4-Flash is a lightweight model meticulously designed by DeepSeek to deliver the ultimate combination of lightning-fast response times and unmatched cost-effectiveness. Engineered with fewer parameters and significantly lower activation overhead, V4-Flash provides an exceptionally fast and economical API service. At its core, V4-Flash demonstrates outstanding reasoning capabilities that closely rival the V4-Pro model. While featuring a slightly streamlined repository of world knowledge, it remains highly capable of satisfying the demands of most application scenarios. In Agentic applications, V4-Flash performs on par with the Pro version when handling standard and fundamental tasks. As the premier choice for developers prioritizing high concurrency, low latency, and cost efficiency, DeepSeek-V4-Flash serves as the optimal solution for deploying large-scale, high-frequency, and lightweight AI workloads.
Deepseek V4 Pro
DeepSeek-V4-Pro is the next-generation flagship open-source large language model developed by DeepSeek, delivering comprehensive performance that rivals the world's premier closed-source models. Compared to its predecessor, V4-Pro achieves a breakthrough evolution in Agentic capabilities. It firmly holds the top position among open-source models in Agentic Coding, providing a high-quality, end-to-end code delivery experience that surpasses mainstream industry benchmarks (such as Sonnet 4.5). Furthermore, the model not only boasts an expansive repository of world knowledge that leads the open-source community, but it also demonstrates ultimate logical reasoning prowess in highly demanding evaluations—including mathematics, STEM, and competitive programming. In these rigorous domains, V4-Pro outperforms all publicly evaluated open-source models and matches the capabilities of global closed-source giants. As the ideal foundational model for building complex agentic workflows, professional-grade software developm
DeepSeek-OCR 2
DeepSeek-OCR 2 is a multimodal document recognition model released by DeepSeek AI, serving as an upgrade to DeepSeek-OCR. By introducing the DeepEncoder V2 architecture, it achieves a paradigm shift in visual encoding from "fixed scanning" to "semantic reasoning." The model replaces the original CLIP encoder with a lightweight language model (Qwen2-0.5B) and incorporates a causal flow query mechanism, while retaining the DeepSeek-3B-MoE decoder. The model requires only 256 to 1,120 visual tokens to cover complex document pages. On the OmniDocBench v1.5 benchmark, it achieves an overall score of 91.09%, a 3.73% improvement over its predecessor, with the reading order recognition edit distance reduced from 0.085 to 0.057.
DeepSeek V3.1
DeepSeek-V3.1 is a hybrid model that supports both thinking mode and non-thinking mode. DeepSeek-V3.1 is fine-tuned on top of DeepSeek-V3.1-Base, which is built upon the original V3 base checkpoint using a two-phase long-context extension approach, following the methodology outlined in the original DeepSeek-V3 report. We have expanded our dataset by collecting additional long documents and substantially extending both training phases. The 32K extension phase has been increased tenfold to 630 billion tokens, while the 128K extension phase has been extended by 3.3 times to 209 billion tokens.
DeepSeek R1 0528
DeepSeek R1 0528 is the latest open-source model released by the DeepSeek team, boasting impressive reasoning capabilities; notably, it achieves performance comparable to OpenAI's o1 model in mathematics, coding, and reasoning tasks.
DeepSeek V3 0324
DeepSeek V3, a 685B-parameter mixture-of-experts model, is the latest iteration of the flagship chat model family from the DeepSeek team.