Starter
For solo builders shipping their first agent.
$29 / month
60 requests / minute
1M tokens / month included, then pay-as-you-go
- All standard agents
- Community support
- Usage dashboard
- 99.9% uptime target
AI agent infrastructure
The dependable pipe between your product and intelligence.
One key plugs your product into production-grade AI agents. Route requests through our switchboard, meter your usage in real time, and scale from prototype to enterprise without touching your stack.
The switchboard
Six production agents, one flat API.
Every agent speaks the same request shape, so switching models is a one-line change. Prices below are the metered rates applied after your monthly allowance.
| Agent | Strength | Metered price | p95 latency | Tags |
|---|---|---|---|---|
| Atlas Reasoner | Complex multi-step reasoning | $0.80 per 1M tokens | 184 ms | Chain-of-thoughtTool use |
| Echo Writer | Long-form marketing and docs copy | $0.35 per 1M tokens | 92 ms | Tone controlSEO mode |
| Scout Coder | Code generation and refactors | $1.20 per 1M tokens | 210 ms | 40+ languagesRepo context |
| Sentry Vision | Image understanding and OCR | $0.60 per 1K images | 340 ms | TablesHandwriting |
| Herald Voice | Speech synthesis and dubbing | $2.40 per 1M chars | 150 ms | 32 voicesStreaming |
| Ledger Ops | Structured data extraction | $0.45 per 1K pages | 260 ms | PDF nativeSchema out |
Metered, honest pricing
Pay for tokens, not for promises.
Every plan includes the full agent catalog. You are never charged for failed requests, retries, or streaming overhead.
Most popular
Scale
For product teams running agents in production.
$99 / month
600 requests / minute
10M tokens / month included, then pay-as-you-go
- All agents + vision
- Priority routing
- 99.95% uptime SLA
- Email support in 24 h
- Webhooks and audit log
Enterprise
For platforms with compliance and scale needs.
from $499 / month
Dedicated throughput
Custom volume with committed-use pricing
- Private model routing
- SOC 2 + DPA + BAA
- 99.99% uptime SLA
- Named engineer on call
- Custom retention
All prices exclude applicable sales tax. Overage is billed per agent at the metered rates above. Full terms.
Integration
From zero to first call in three lines.
Standard REST over HTTPS with JSON in and JSON out. Official client libraries for TypeScript, Python and Go ship with retries, streaming and typed tool-use built in.
- 01
Create your key
Sign up and mint a scoped key in the dashboard. Keys are shown once and stored hashed.
- 02
Point the client at us
Set the base URL and key. The agent catalog is one path parameter away.
- 03
Ship it
Add up to 10 keys per project, each with its own spend cap and route allowlist.
# pip install nexapi from nexapi import Client client = Client(api_key="sk_live_•••") response = client.agents.run( agent="atlas-reasoner", input="Summarize this quarter's churn drivers from the warehouse.", ) print(response.output)
99.95%measured uptime, last 12 months
180 msmedian latency, US-East region
30 daysdefault log retention, zero on request
SOC 2Type II audited, report on request
Questions from buyers
The fine print, up front.
+What happens when I exceed my monthly token allowance?
Nothing breaks. Overage bills at the same per-token rates listed in the agent catalog, and you can set a hard spend cap per key in the dashboard at any time.
+Can I try NexAPI before paying?
Yes. Every new account gets 50,000 free tokens and full access to the agent catalog for 14 days, no card required.
+Do you train models on my requests?
No. Request and response bodies are used only to serve your call. Nothing is retained beyond 30 days of operational logs, and Enterprise plans can set retention to zero.
+How do I get billed?
Plans bill monthly on the card on file. Metered overage is invoiced at the start of each cycle with a full per-agent breakdown, and everything is exportable via the API.
+Is there a service credit if you miss the SLA?
Yes. Scale and Enterprise plans include automatic credits: 10% of the monthly plan for every 0.1% of measured downtime below target, up to 100%.