Inference
Unified gateway to Claude, GPT, Gemini, Grok, Llama, Mistral and more. One API, every frontier model — with credit-based metering built in.
Production ready infra for AI. Gateway, Search, RAG and Sandboxing out of the box.
Unified gateway to Claude, GPT, Gemini, Grok, Llama, Mistral and more. One API, every frontier model — with credit-based metering built in.
Isolated execution environments spun up per session. Run untrusted code safely, stream results, and tear down instantly.
Retrieval-augmented generation on your own data. Chunk, embed, and query documents — grounded answers without hallucination.
Live web search piped directly into your model context. Real-time results, auto-cited — no prompt engineering required.
import { Agent } from 'easysandbox';// Initialise the client with your API tokenconst agent = new Agent(process.env.NOVISURF_API_TOKEN);// Start a session — sandbox + search tools, streaming enabledconst session = await agent.start({model: 'novisurf/oversurf-swift-1',sandbox: true,tools: ['sandbox', 'search'],systemPrompt: 'You are a helpful coding assistant.',});// Stream the response chunk-by-chunkconst stream = session.stream('Write and run a Python script that fetches live BTC price.');for await (const chunk of stream.textStream) {process.stdout.write(chunk);}// Await final result + token usageconst { text, usage } = await stream.done();console.log(`Cost: $${usage.cost_usd.toFixed(6)}`);// Tear down the sandbox when doneawait session.terminate();
Pick from 20+ models across providers like OpenAI, Anthropic, Google, Mistral, and more — all through a single unified API.
Use our SDK to add chat, search, sandboxed code execution, and real-time streaming in minutes, not weeks.
We handle rate limiting, caching, observability, and failover — so you can focus on building, not infrastructure.
Perfect for getting started and testing our platform.
For developers shipping production agents.