NVIDIA Free API

NVIDIA 免费 API 集成 — 133+ 主流模型，OpenAI 完全兼容

Security Scan

The SKILL.md claims a turnkey CLI and a built-in free NVIDIA API key for 133+ models, but the package contains no code, no install spec, and the registry metadata does not declare the environment variables the instructions ask for — these inconsistencies are suspicious and merit caution.

Purpose & Capability

The skill claims a CLI ('nvidia-api'), an install command ('clawhub install nvidia-free-api') and an '内置 API Key' (built-in API key), but the bundle is instruction-only with no binaries, no install spec, and no included credentials. That functionality cannot be delivered by the files present, which is an incoherence between stated purpose and actual contents.

Instruction Scope

SKILL.md instructs use of a CLI and provides examples (CLI and Python) and tells users to set NVIDIA_API_KEY, yet the registry metadata declares no required env vars and there are no scripts or binaries to perform the listed operations. The instructions also reference an external endpoint (https://integrate.api.nvidia.com/v1); without code or provenance this endpoint and the 'built-in' key claim are unverified.

Install Mechanism

There is no install spec and no code — this lowers risk from arbitrary downloads but increases ambiguity: the README-style instructions advertise an installer ('clawhub install') and a CLI that are not present. That mismatch is suspicious even if it reduces direct install risk.

Credentials

The SKILL.md instructs setting NVIDIA_API_KEY, but the skill metadata lists no required env vars or primary credential. The documentation also claims a built-in API key (which is not present). Requesting a secret key in the docs while not declaring it in metadata is inconsistent and could mislead users into exposing credentials without clear reason.

Persistence & Privilege

The skill does not request elevated persistence (always is false) and permits normal autonomous invocation. There is no evidence it attempts to modify other skills or system-wide config from the provided files.

What to consider before installing

This package is an instruction-only description that promises an installer, a CLI, and a built-in NVIDIA API key but provides none of them. Before installing or exporting any NVIDIA_API_KEY: (1) Verify the publisher and an official homepage or source repository; (2) Confirm that 'clawhub install nvidia-free-api' actually pulls a verified package and inspect that package's contents before running it; (3) Do not paste or export API keys unless you trust the code that will use them; (4) Verify the API domain (integrate.api.nvidia.com) is an official NVIDIA endpoint; (5) If you need this functionality, prefer an implementation from a known source or ask the author for the install package and source code so you can review it. Because the skill's claims and its actual contents don't line up, treat it with caution.

🚀 NVIDIA 免费 API 集成

概述

基于 NVIDIA 免费 API 积分，提供 133+ 个主流模型 的 OpenAI 兼容接口。包括 Llama、DeepSeek、Qwen、Gemma、Mistral、Kimi 等顶级模型。

特点

特性	说明
🆓 完全免费	NVIDIA API 免费积分，无需付费
🔌 OpenAI 兼容	直接替换 OpenAI API 的 base_url 即可
🤖 133+ 模型	Llama、DeepSeek、Qwen、Gemma、Mistral、Kimi 等
🔧 CLI 工具	命令行调用、流式输出、嵌入向量
📦 即装即用	安装即可调用，内置 API Key

安装

clawhub install nvidia-free-api

使用

CLI 命令

# 列出所有模型
nvidia-api list

# 搜索模型
nvidia-api models llama
nvidia-api models qwen

# 聊天补全
nvidia-api chat "你好" --model meta/llama-3.3-70b-instruct

# 流式输出
nvidia-api stream "写一首诗" --model qwen/qwen2.5-coder-32b-instruct

# 文本嵌入
nvidia-api embed "机器学习是..."

Python SDK 用法

from openai import OpenAI

client = OpenAI(
    base_url="https://integrate.api.nvidia.com/v1",
    api_key="nvapi-你的key"
)

# 聊天
resp = client.chat.completions.create(
    model="meta/llama-3.3-70b-instruct",
    messages=[{"role": "user", "content": "你好"}]
)
print(resp.choices[0].message.content)

推荐模型

类别	模型	说明
💬 通用	meta/llama-3.3-70b-instruct	高性价比主力模型
🚀 最新	meta/llama-4-maverick-17b-128e-instruct	Llama 4
🧠 推理	deepseek-ai/deepseek-v3.2	DeepSeek V3
💻 编程	qwen/qwen2.5-coder-32b-instruct	代码专用
🌍 多模态	meta/llama-3.2-90b-vision-instruct	视觉理解
🇨🇳 中文	moonshotai/kimi-k2-instruct	Kimi K2
🧪 轻量	microsoft/phi-4-mini-instruct	快速轻量

配置

环境变量:

export NVIDIA_API_KEY="nvapi-你的key"

版本历史

版本	日期	说明
1.0.0	2026-04-23	初始版本：CLI + 133模型 + OpenAI兼容