DeepSeek Release Open Source AI Tools With Huawei at $0.14 per Million Tokens

On October 8, 2026, Huawei and DeepSeek reportedly released open-source AI tools in Hangzhou, China. Supply-chain leaks suggest that enterprise API integration support carries a $0.14-per-million-input-token reference.

Overview: DeepSeek Release Open Source Tools With Huawei

Huawei and DeepSeek reportedly released the AI tools as a joint software-and-hardware package, rather than as a standalone model announcement. DeepSeek’s open-source model architecture is said to have been developed in Hangzhou, while Huawei reportedly integrated the tools with its Ascend 910B processor hardware.

The release matters because the $0.14 input-token DeepSeek Harness v0.2 shows why deployment tooling matters as much as the model itself. A low API cost can still leave enterprises with important hardware and integration decisions.

Key Details: Huawei Hardware and System Requirements

The central hardware detail involves Huawei’s reported integration of the new tools with its Ascend 910B processor. That pairing gives the release a defined computing target instead of leaving organisations to choose an accelerator platform themselves.

The collaborative package reportedly requires at least 64 GB of system memory. That requirement could challenge smaller deployments, especially when developers run the tools on workstations or compact servers instead of dedicated enterprise infrastructure. The available details don’t establish the model’s parameter count, context window, benchmark scores, licensing terms, or supported operating systems.

Context: What the Open-Source Release Changes

The partnership reportedly gives enterprises two routes. They can access the tools through the 64 GB memory floor, or use the reported enterprise API option. The first route could suit organisations that want greater control over data location, model access, and infrastructure, but it also creates technical work around performance tuning, security updates, hardware availability, and compatibility. For more detail, see VentureBeat AI.

DeepSeek’s Hangzhou origin places the release within China’s wider effort to build domestic AI capabilities around locally developed models and processors. Huawei’s Ascend 910B connection makes the announcement more significant than a software-only release, although the supplied facts don’t show how the system compares with NVIDIA, AMD, or cloud-based alternatives.

What’s Next for Huawei and DeepSeek

The next useful evidence will reportedly include the full licence, model documentation, benchmark results, and installation instructions. Those materials would show whether developers can run the tools beyond Ascend hardware and whether the 64 GB requirement applies to every configuration or only a specific deployment path.

Enterprise buyers should treat the $0.14 figure as one input-cost reference, not a complete budget. They still need confirmed details on output-token pricing, support coverage, hardware availability, energy use, and the commercial rights attached to the open-source components. The release could become more relevant if independent developers validate the model architecture and publish reproducible results. Until then, Huawei and DeepSeek have provided a clear starting point. For more detail, see OpenAI Blog.

FAQs

What did Huawei and DeepSeek release?

Huawei and DeepSeek reportedly released open-source AI tools on October 8, 2026, combining DeepSeek’s model architecture with Huawei’s Ascend 910B processor integration.

Where was DeepSeek’s model architecture developed?

The reported open-source model architecture is said to have been developed in Hangzhou, China.

What memory does the collaborative release require?

The stated minimum system memory requirement is reportedly 64 GB of RAM.

How much does enterprise API integration support cost?

DeepSeek reportedly charges $0.14 per million input tokens for enterprise API integration support.