Free! Luchen joins hands with Huawei Ascend to launch DeepSeek R1 inference API and cloud image service

Written by

Jasper Cole

Updated on:July-17th-2025

The open source of the DeepSeek-R1 series of models has sparked heated discussions and attention around the world due to its outstanding performance and low development costs.

Luchen Technology has teamed up with Huawei Ascend to jointly release the DeepSeek R1 series inference API and cloud image service based on the domestic 910B computing power.

DeepSeek R1 Series Inference API Free Trial

In this release, Luchen used its self-developed domestic inference engine as the technical foundation and successfully achieved the inference adaptation optimization of the domestic Huawei Ascend 910B computing power and DeepSeek R1 series models. The performance is on par with using high-end GPUs, providing developers with efficient, flexible and stable AI inference services, helping enterprises to reduce costs and increase efficiency, and accelerate the rapid deployment of intelligent businesses based on domestic software and hardware systems.

Empowered by domestic computing power, performance is comparable to high-end GPUs

Breakthrough adaptation to the domestic Huawei Ascend computing power: Luchen's self-developed inference engine has successfully achieved deep adaptation and optimization with the domestic Huawei Ascend computing power, benchmarking the inference performance of the industry's high-end GPU resources, contributing to the construction of the domestic AI ecosystem.

Luchen also supports mainstream computing power such as NV H800: In addition to domestic computing power, DeepSeek R1 series reasoning API also supports common computing power such as NV H800, fully unleashing the potential of heterogeneous clusters and meeting reasoning needs in different scenarios.

Reasoning Demonstration

Flexible model, full coverage of scenarios

Luchen Cloud Platform provides a variety of flexible options for DeepSeek R1 series models, from full-blooded 671B large models to efficient distillation small models, to meet the diverse needs of developers and enterprises. Users can accurately select models based on business scenarios to achieve the best balance between accuracy and efficiency.

Stable API service, free and ready to use

To meet the needs of commercial deployment, Luchen DeepSeek R1 series inference API provides stable and flexible API service capabilities to ensure that the model runs smoothly in a large-scale production environment, helping enterprises to quickly implement AI applications. In addition, thanks to the simple and easy-to-use interface, developers can quickly integrate without paying attention to the underlying details and focus on business innovation.

Currently, the DeepSeek R1 series inference API is open for unlimited free trial for a limited time, and we sincerely invite developers to experience it first.

Lu Chenyun Reasoning Mirror

For users with high customization requirements and the need to deploy DeepSeek R1 series models privately, Luchen Technology also provides inference images based on open source solutions through Luchen Cloud, which can easily start Ascend 910B or NV H800 computing resources and corresponding images to deploy private inference services. It also supports a variety of computing powers from full-blooded 671B to distilled small models, domestic Ascend and NV H800, etc., and users can flexibly choose.

Reasoning Demonstration

For example, after logging in to Luchen Cloud, select the "ASCEND-910B" instance, use the DeepSeek-R1-Distill-Qwen-7B model and image, and reason about the question: How many positive whole-number divisors does 196 have? Put answer in \box {Your answer} (How many positive whole-number divisors does 196 have?), and from the corresponding output, you can see that the model has thought deeply and found the correct answer. By selecting the corresponding instance and image, you can further privately deploy the full-blooded 671B DeepSeek R1 model.

Luchen Video Ocean Vincent Video Free Trial

Luchen Video Ocean V2.0 is officially here. This upgrade is not only a "big step evolution" in terms of model, speed, and functional gameplay, but it is also completely free and waiting for you to grab the wool. The innovative technical architecture and user-oriented video generation experience present you with a brand new digital interactive world.