One platform.
Every LLM capability.
The complete managed gateway: 20+ models, smart routing, built-in guardrails, real-time cost tracking, prompt management, and team controls — every feature on every plan, no provider keys to manage.
No credit card · 1 provider live · Anthropic + OpenAI shipping next
- 20+Models liveGoogle Vertex AI
- <2msGuardrail overheadP50, in-process
- 0%Feature gatingEvery plan, every feature
- 4 → 0%Platform feeOpenRouter charges 5%
One key. Every model. Zero provider config.
Drop-in OpenAI-compatible API. Change two lines of code and you can call any model in the catalog. We manage every provider key, balance load across deployments, and fall over automatically when a provider degrades.
- 20+ models live now across 1 provider — Anthropic, OpenAI, AWS Bedrock shipping next
- Usage-based + latency-based routing across deployments per model
- Zero-config failover, weighted load balancing, model aliases
Block abuse, redact PII, stop prompt injection — by default.
Five guardrails active on every request out of the box. Pre-, during-, and post-call hooks. Sub-2ms overhead because they run in-process, not over the network.
- PII redaction (5 categories: email, phone, SSN, credit card, IP)
- Prompt-injection detection on every inbound request
- Per-key, per-team, or per-org scope — override at call time via metadata.guardrails
Real-time spend per model, team, key — settled to the cent.
Every request returns the router-computed cost in the response header. We settle credits atomically against your balance. Nothing hidden, nothing rounded, nothing inferred from token math.
- Per-model breakdown with daily aggregates and dimensional drill-down
- Reserve + settle pattern means no negative balances, ever
- Webhook + CSV export for finance teams
Spending caps at every layer — automatic enforcement.
Hard and soft limits per organization, team, or individual API key. Budgets reset on your cycle (daily, weekly, monthly). When a key trips a hard limit, it returns 402 — no further spend possible.
- Per-org, per-team, or per-key — composable limits, smallest scope wins
- Daily / weekly / monthly reset cycles with budget_reset_at timestamps
- Slack + webhook alerts at 70% / 90% / 100% utilization
The rest of the platform.
Every capability above is one half of the gateway. The other half is observability, prompts, teams, and the tools to test it all interactively.
Analytics
Per-key, per-team, per-model spend reports with drill-down by hour or tag.
Observability
Request logs, latency percentiles, error rates — pushed to Langfuse, Datadog, or S3.
Prompts
Versioned templates, A/B variants, Jinja2 variables — all server-side.
Teams
Org → Team → Key hierarchy with role-based access and per-team budgets.
Playground
Test any model interactively in the browser. Sticky API key, streaming responses.
One key. One bill. Zero config.
Buy credits, call the API. We handle routing, cost tracking, guardrails, and safety. You never see a provider API key.
Buy credits
Stripe checkout
Get API key
sk-nemo-xxx
Call models
OpenAI SDK
We route
20+ models
You build
Ship to production
Ship with one key.
Without managing any.
Every feature on every plan. No provider config, no feature paywalls, no infrastructure to operate. Your first request is routed, guarded, and logged automatically.
$5 in credit on signup · No credit card required
Start building today
One API key.
Every model. Three tiers.
4% pay-as-you-go, 2% on Tier 2 monthly, 0% on Tier 3 annual. All enterprise features included on every tier.
No credit card required · OpenAI-compatible API
AI Models
from leading providers
Providers
Google Vertex AI
Platform Fee
on Tier 3 annual plans
Uptime SLA
enterprise-grade reliability