New: Gemini 2.5 Flash & Remote Resources

Power Industrial & Commercial Inference with NeuralHub

The unified AI Gateway for logistics, manufacturing, and commercial scale. Integrate top-tier models and execute code securely in the cloud.

No credit card required
Generous free tier
neuralhub-cli — v4.5.1
➜~neuralhub models list
gemini-2-flashAvailable
claude-3.5-sonnetAvailable
➜~neuralhub resources create --lang python
Initializing MicroVM 'sbx-py-7a2'... Ready (300ms)
Supported Providers
Google
Anthropic
OpenAI
Meta
Mistral
Daytona

How NeuralHub Works

A streamlined workflow to get your autonomous agents from concept to production in minutes.

1. Provision Resources

Spin up isolated, stateful MicroVMs in milliseconds. Give your agents a secure environment to execute code and store files.

2. Attach MCP Tools

Equip your agents with capabilities from our Skills Marketplace. Connect to GitHub, databases, or web scrapers instantly.

3. Deploy Agents

Select your preferred LLM (Claude, Gemini, OpenAI) and deploy. Monitor activity, logs, and memory in real-time via the dashboard.

The Complete AI Stack

Stop stitching together API keys and invoices. NeuralHub provides the complete stack for autonomy.

Universal LLM API

One interface for OpenAI, Anthropic, Gemini, and 100+ open source models. Switch models with one line of code.

Serverless Resources

Execute untrusted code securely in millisecond-startup MicroVMs. Perfect for code interpreters and agents.

Persistent Memory

Integrated vector and object storage for your agents. Give them long-term memory out of the box.

MCP Tool Ecosystem

Standardized tools for web browsing, data extraction, and API interactions via Model Context Protocol.

Unified CLI & SDK

Manage your entire infrastructure from the terminal. TypeScript, Python, and Go SDKs available.

Built for Developers

NeuralHub's SDK makes it incredibly simple to spin up isolated environments, attach MCP tools, and deploy autonomous agents with just a few lines of code.

  • Type-safe SDKsAvailable for TypeScript, Python, and Go.
  • Drop-in IntegrationsWorks seamlessly with Vercel AI SDK and LangChain.
  • Stateful ExecutionPause, resume, and fork agent states on the fly.

Transparent, Usage-Based Pricing

No monthly subscription fees for access. You only pay for the compute and tokens you consume.

Model Inference
Pass-through pricing + small platform fee
Resources
Starting at $0.0001/sec
View credit packages
Estimated Costs
Typical monthly spend for active agents
Python 3.10
MicroVM
$0.0001/sec
Available
Node.js 20

Ready to build the future?

Join thousands of developers building the next generation of autonomous agents on NeuralHub.

Product
  • Models
  • Resources
  • Storage
  • Credits
Resources
  • Documentation
  • API Reference
  • Guides
  • Status
Company
  • About Us
  • Blog
  • Careers
  • Privacy & Terms
Connect
© 2026 NeuralHub Inc. All rights reserved.

Durable Workflows

Build fault-tolerant agent loops that survive server restarts and timeouts. Powered by Upstash.

agent.ts
import { NeuralHub } from '@neuralhub/sdk';

// Initialize the client
const nh = new NeuralHub(process.env.NH_API_KEY);

async function runAgent() {
// 1. Create an isolated MicroVM
const env = await nh.environments.create({
runtime: 'python-3.10',
tier: 'micro'
});

// 2. Deploy an agent with MCP tools
const agent = await nh.agents.deploy({
model: 'claude-3.5-sonnet',
environmentId: env.id,
tools: ['mcp://github', 'mcp://firecrawl']
});

return agent.execute("Analyze the repo and write tests.");
}
MicroVM
$0.0001/sec
Available
Firecrawl
Web Scraper
$0.005/page
Available