算力
面向 AI Agent 的隔离代码执行沙箱：在容器中安全运行不可信代码，支持 GPU 算力。以 USDC 按次计费，无需订阅。
面向 AI Agent 的隔离代码执行沙箱：在容器中安全运行不可信代码，支持 GPU 算力。以 USDC 按次计费，无需订阅。
下列每个端点都可以用同一个 API 密钥调用，或通过 x402 以 USDC 按次付费。没有订阅，也没有最低消费，用多少付多少。
Modal Sandbox
Isolated code execution in secure containers
Secure Python 3.11 runtime for AI agents. Create isolated environments (1 vCPU, 1 GiB RAM, up to 5-min burst or 24-hour sessions). Execute commands, capture stdout/stderr/exit codes, persist state between calls for multi-step workflows. Use cases: code interpretation, package verification, data processing, automation scripts. Works with Chat Completions (function calling) to let AI models run code safely. No account needed — pay per call via x402.
定价: From $0.0011/call
| 付款方式 | 端点 | 价格 | 说明信息 |
|---|---|---|---|
POST | /sandbox/create | $0.0121 | Create isolated sandbox |
POST | /sandbox/exec | $0.0011 | Execute command in sandbox |
POST | /sandbox/status | $0.0011 | Check sandbox state |
POST | /sandbox/terminate | $0.0011 | Destroy sandbox |
如何付费
直接调用任意端点。不带密钥时会收到 HTTP 402，其中写明精确的 USDC 价格；用 Base 或 Solana 钱包签名该报价后重发请求即可。使用账户 API 密钥时，同一请求改为从余额扣费。
curl -i https://api.jarvisclaw.ai/v1/chat/completions \
-H "Content-Type: application/json" \
-d '{"model":"auto/free","messages":[{"role":"user","content":"Hello"}]}'