You lock down every other part of your stack with zero-trust. Your AI shouldn't be the exception. Now you can deploy trust checks three ways — MCP, Python, or Node — free, on your own machine, with no API key. No keys. No cost. No excuse. Here's the whole menu, kept simple.
Trust shouldn't be gatekept
For years, answering one simple question about an AI system — is it safe, fair, and compliant? — was a privilege. It meant hiring an auditor, waiting weeks, and writing a cheque that often ran into the tens of thousands. A single hiring-bias audit alone can cost $50,000. That price tag put trust out of reach for exactly the people who could do something about it: the students, solo builders, startups, and engineers actually shipping the AI.
We think that's backwards. So we did the opposite of an enterprise sales motion: we made TrustModel free, open, and one line away — and we put it wherever you already build.
Three free ways to deploy — no key required
You don't need to weigh a menu of options. There are three toolkits — MCP, Python, and Node — and each one has a free version you run right now, on your own machine, with no API key. Pick the one that matches how you work. That's the whole decision.
All three run 100% on your machine — no sign-up, no key, nothing sent to us. Private by design, and genuinely, forever free. Try it in 30 seconds, run it in CI, or work fully air-gapped.
When you're ready for calibrated, audit-ready scores you can hand to a regulator, each toolkit has a hosted cloud upgrade too — free to start (5 credits / $500 of evaluation, no credit card). But you never need it to begin, and you're never re-learning anything: same tools, same names.
The three free ways, at a glance
Copy one line and you're running. The free command is all you need to start — the hosted upgrade is there for later.
MCP — AI assistants and agents
Free — no key: npx
Hosted upgrade:
Python
Free — no key: pip install trustmodel-local (
Hosted upgrade: pip install trustmodel (
Node / TypeScript
Free — no key: npm i
Hosted upgrade: npm i
"Okay, but which one is me?"
Use MCP if you live in an AI assistant. If you build in code, choose the matching Python or Node / TypeScript toolkit. Start local; add hosted when you need a score you can defend.
Democratized, on purpose
The commercial AI-governance world is built for a Chief Compliance Officer with a six-figure budget — months away from the person who actually ships the model. We flipped it. The whole local toolkit is free and open source; the calibrated, audit-ready scores and compliance reports are the paid layer for when you need to prove it to a regulator. Think Linux and Red Hat: run the free version forever, pay only when the stakes demand it. We'd rather a million builders check their AI for free than a thousand do it after the lawsuit.
Free gets you building. A key gets you audit-ready.
The free, local version scores your AI using your own LLM as the judge — ideal for building, iterating, and gating CI. But the day someone asks you to prove it's safe and compliant — an auditor, a regulator, your board, a big customer's security review, or a plaintiff's attorney — "our own model scored it well" isn't an answer. That's when you reach for a key. The same tools then become a defensible, enterprise-grade assurance platform:
A calibrated, audit-ready TrustScore — a 0–100 rating that's benchmarked and consistent, not one model's opinion. The number you can put in a report and defend.
Compliance reports mapped to 30+ regulations — EU AI Act, NIST AI RMF, ISO 42001, NYC Local Law 144, OWASP LLM Top 10, HIPAA, ECOA and more — the evidence auditors actually accept.
Continuous monitoring in production — dashboards, drift and threshold alerts, OpenTelemetry export. Catch problems the day they start, not the quarter they're subpoenaed.
Runtime agent governance — allow / deny / redact every risky action before it reaches a user, with an in-VPC Edge so data never leaves your network.
AgentCert & certification — certify an agent and publish a verifiable badge your partners and customers can check.
Enterprise-ready — SOC 2, RBAC and SSO, and multi-cloud (GCP, Azure, AWS).
Getting there takes about a minute: sign up, get 5 free credits — $500 of evaluation, no credit card — and drop the key into the same toolkit you're already using.
Pick your toolkit. Trust your AI in 30 seconds. Free.
No key and no signup for the local versions. A free key with $500 of credits when you're ready for the cloud. for more in-depth discussion and hearing/ meeting the top leaders in Ai on this topic with hands-on labs, register your interest to attend our summit at Stanford University on October 1st, 2026 here: [https://trustmodel.ai/summit2026]
Top comments (0)