HPC-AI OpenClaw Provider

HPC-AI provider plugin for OpenClaw. It uses the OpenAI-compatible chat completions API at https://api.hpc-ai.com/inference/v1.

Install

After the package is published on ClawHub:

openclaw plugins install clawhub:@lioz129/openclaw-hpc-ai-provider

For a Git checkout during development:

git clone https://github.com/lioZ129/openclaw-hpc-ai-provider.git
cd openclaw-hpc-ai-provider
pnpm install
pnpm check

Configure

Set your HPC-AI API key:

export HPC_AI_API_KEY="your-hpc-ai-api-key"

Optional base URL override:

export HPC_AI_BASE_URL="https://api.hpc-ai.com/inference/v1"

Default model ref:

hpc-ai/deepseek/deepseek-v4-flash

OpenClaw model refs include the provider id first. The upstream HPC-AI model id for the default model is deepseek/deepseek-v4-flash.

Models

Model ref	Input	Context	Max output	Input	Output	Cache read
hpc-ai/minimax/minimax-m2.5	text	196,000	195,000	0.30	1.20	0.03
hpc-ai/moonshotai/kimi-k2.5	text, image	256,000	256,000	0.60	3.00	0.10
hpc-ai/deepseek/deepseek-v4-pro	text	1,048,576	128,000	1.74	3.48	0.145
hpc-ai/deepseek/deepseek-v4-flash	text	1,048,576	128,000	0.14	0.28	0.028
hpc-ai/moonshotai/kimi-k2.6	text, image	256,000	256,000	0.95	4.00	0.16
hpc-ai/zai-org/glm-5.1	text	202,000	202,000	1.40	4.40	0.26