IBM announced that its AI development platform watsonx.ai will add new Llama3.18B and Llama3.370B models that support the DeepSeek-R1 distilled version. DeepSeek uses knowledge distillation technology to optimize multiple Llama and Qwen model variants, significantly improving model performance. This move injects strong inference capabilities into the watsonx.ai platform, expands its model ecosystem, and provides enterprises and developers with more convenient and efficient AI solutions.
On the watsonx.ai platform, users can directly deploy Llama distillation model through the “Deploy on Demand” directory, or upload other DeepSeek-R1 variants, such as the Qwen distillation model, through the “Custom Base Model” import function, to meet various Application requirements. The powerful reasoning capabilities of the DeepSeek-R1 model make it suitable for a wide range of fields, providing users with flexible and efficient AI services.
DeepSeek-R1 has powerful reasoning capabilities and is suitable for a wide range of fields, providing enterprises and developers with efficient and flexible AI solutions. This update further enriches the model ecosystem of watsonx.ai, helping users to develop and deploy AI applications more conveniently.
The update of watsonx.ai marks significant progress in AI model optimization and platform construction, providing users with stronger AI tools and a more convenient user experience, further consolidating its competition in the AI field force.