1NCE AI
17+ Models, One API Key, Live in 3 Minutes
///
)
Powered by AWS and models from leading providers
)
){kind=logo}
){kind=logo}
){kind=logo}
Connect to Claude, Llama, Mistral, Amazon Nova and more through a single OpenAI-compatible endpoint. At standard AWS Bedrock pricing, with no AWS account, IAM setup, or per-model approval required. You're live in under 3 minutes.
1NCE AI Insights:
17+ hosted models
One single API Key
$5 free credits to start
No AWS account required
< 3 min to first API call
99.9% target availability
A Smarter Way to Run AI at Scale
An AI platform built for IoT teams, enterprise developers, and platform engineers building across Asia-Pacific.
One Key, Every Model
Access all 17+ models with a single API key. No separate provider accounts or IAM policies to manage.
OpenAI-Compatible API
Works out of the box with LangChain, LlamaIndex, Vercel AI SDK, LiteLLM, and any OpenAI-compatible tool. No new SDK.
Spend Control and Governance
Apply rate limits and spending caps across teams and models. Monitor usage with complete audit logs and predictable costs.
Data Privacy by Default
Never used to train models. Optional per-key PII redaction, no code changes.
One Invoice for Everything
Consolidate usage across models into a single invoice.
Set budgets, track spend in real time.
IoT-Native Integration
Current 1NCE customers can add AI through the same portal.
Same invoice, same support.
Zero Data Training Required
Prompts and outputs don't train foundation models. Contractual for Bedrock-hosted; US-routed follows provider terms.
An AI Platform Created for Responsible AI
Production-ready infrastructure with the controls enterprise teams expect:
Self-Service Portal
Create accounts, generate API keys, configure budgets, and manage access from one single AI platform.
Usage Dashboard
Monitor consumption by model, key, and team, set budget alerts, and export billing reports.
Full Audit Logging
Every API call includes model information, token counts, latency metrics, and costs.
Multi-Model Routing
Automatically switch to an alternative model if the selected model becomes unavailable.
)
Get Started in Four Steps
Developer-First Setup
Create Your Account
Sign up with your work email. No AWS account, IAM setup, or model access requests needed. Existing 1NCE customers can use their current login.
Generate an API Key
Choose your models, define optional spending limits, and create a scoped API key.
Make Your First Call
Point your OpenAI-compatible client at api.1nce.ai/v1. Drop-in compatible, without new SDKs or refactoring.
Track Usage and Billing
Track usage by model, API key, and billing period, with full transparency and no minimum spend.
Pricing
Pay Only for What You Use
Pricing is usage-based, at standard AWS Bedrock rates for tokens consumed. There's no minimum spend and no hidden fees. Start with $5 in free credits and top up as needed.
What Teams Build with 1NCE AI Platform
AI Features Inside Your Product
Add AI capabilities to your SaaS platform without managing multiple model providers and accounts.
Internal AI Chat
Deploy OpenWebUI or LibreChat internally, with SSO, audit logs, and enterprise-grade user controls.
IoT Data Interpretation
Convert telemetry into plain-language insights, anomaly summaries, and operational reports.
Automated Document Processing
Summarise contracts, extract structured data, and review compliance documents with a single API call.
Start Building with 1NCE AI in Asia-Pacific
Your first AI-powered application in minutes. Sign up and generate one single API key for every model.
FAQ
I already call OpenAI directly. How long does switching actually take?
I already call OpenAI directly. How long does switching actually take?
Two lines of code: change base_url to https://api.1nce.ai/v1 and swap your API key. That's it. The endpoint uses the same OpenAI request and response format, so your existing SDK, prompts, and parsing logic stay unchanged.
Most developers make their first successful call within a minute of generating a key. The model name changes (e.g. "claude-3-5-sonnet" instead of an OpenAI model ID), but the call structure is identical.
How is this different from signing up for AWS Bedrock directly?
How is this different from signing up for AWS Bedrock directly?
With Bedrock you need a full AWS account, IAM setup, and individual access requests for each model. There's also no built-in team spend management or audit dashboard. With 1NCE AI you get all 17+ models on day one through an OpenAI-compatible API, a real-time spend dashboard with per-key caps, full audit logs, and AWS Bedrock list pricing with no markup. No AWS account required.
We use LangChain / LlamaIndex / Vercel AI SDK. Does it work?
We use LangChain / LlamaIndex / Vercel AI SDK. Does it work?
Yes. Any client or framework that supports a custom base_url and OpenAI-compatible completions format works with 1NCE AI. This includes LangChain, LlamaIndex, Vercel AI SDK, LiteLLM, and any other tool that lets you configure the endpoint and API key separately. If it works with OpenAI, it works with 1NCE AI.
Can separate teams or projects use their own keys and budgets?
Can separate teams or projects use their own keys and budgets?
Yes. You can create as many scoped API keys as you need, one per project, per team, per environment (dev/staging/prod), or per customer if you're building a multi-tenant product. Each key has its own spend cap, and usage is tracked individually in the dashboard. Revoke any key instantly without affecting the others.
Is there a free tier, or do we have to commit to a paid plan to evaluate it?
Is there a free tier, or do we have to commit to a paid plan to evaluate it?
No commitment required. Every new account gets $5 in free credits at sign-up — enough to make thousands of API calls and evaluate the platform across multiple models. No credit card required to start. If you want to continue, you top up whenever you're ready. There's no subscription, no monthly minimum, and no time limit on using your credits.
What happens if a model goes down mid-request?
What happens if a model goes down mid-request?
1NCE AI includes automatic multi-model routing. If the primary model for a request becomes unavailable, the platform routes to the next best alternative without any action needed on your end and without downtime. You still get a response; you don't need to build your own fallback logic.
We're running an IoT platform with high call volumes. Can 1NCE AI handle that?
We're running an IoT platform with high call volumes. Can 1NCE AI handle that?
Yes, IoT-scale workloads are a primary use case. 1NCE AI runs on AWS Bedrock infrastructure with a 99.9% target availability and low-latency endpoints specifically optimized for telemetry analysis, predictive maintenance, and device data interpretation. Rate limits are configurable per key so high-volume keys can be allocated appropriate headroom. If you're running large-scale IoT workloads, talk to the sales team: there are volume arrangements available for predictable high-consumption accounts.
Do you support streaming responses?
Do you support streaming responses?
Yes. The 1NCE AI endpoint supports streaming in the same format as the OpenAI API. Set stream: true in your request and handle the server-sent events the same way you would with any OpenAI-compatible client. No changes needed if you're already handling streaming in your application.
Newsletter