Apps
What teams build on nRouter.
Reference architectures for coding assistants, customer support, RAG pipelines, document extraction, voice agents, and compliance screening.
Reference architectures.
Common patterns on one OpenAI-compatible API, with routing, guardrails, and budgets in the request path.
Code and refactor assistant.
Send quick completions to fast models and multi-file refactors to frontier models, one string apart.
- Fallback models you choose, tried in order on a provider fault
- Secret scanning and prompt-injection defense in the request path
- Per-key budgets for every developer or CI job
Customer support copilot.
A multi-turn support agent for live chat and ticket triage, with PII redacted before dispatch.
- PII redaction and content moderation before a provider is called
- Per-team and per-key budget enforcement
- Request logs with model, tokens, and cost for every reply
Knowledge and RAG engine.
Retrieval and synthesis that grounds answers in your own documents.
- Embeddings and long-context chat models behind one key
- Response caching for repeated questions
- Prompt-injection screening on retrieved context
Contract and financial document analyzer.
Long-context extraction that turns filings and contracts into structured JSON.
- Structured JSON output on models that support it
- Per-request cost on every response
- Audit logs for who changed keys, teams, and budgets
Voice agent.
Transcription, speech synthesis, and conversational turns under one credit balance.
- Audio and chat endpoints under one virtual key
- Credits reserved before each call, settled at list price
- Spend tracked per key and per team
Fraud and compliance screener.
Transaction and policy review with PII redaction and role-based access to every key.
- PII redaction before a prompt reaches a provider
- Role-based access across org, team, and key
- Hard spend ceilings per key
One SDK call across every workload.
Every reference architecture starts from the same client and the same canonical SDK snippets.
pip install nrouter-sdk| 1 | # Cache: enabled (org default). Pass nrouter_cache: false to skip. |
| 2 | from nroutersdk import nRouter, nRouterGuardrailBlockedError |
| 3 |
|
| 4 | client = nRouter() # reads NROUTER_API_KEY; defaults to https://api.nrouter.ai/v1 |
| 5 |
|
| 6 | try: |
| 7 | response = client.chat.completions.create( |
| 8 | model="gpt-5.4-mini", |
| 9 | temperature=1, |
| 10 | max_completion_tokens=1024, |
| 11 | messages=[ |
| 12 | {"role": "user", "content": "Hello! What models do you support?"}, |
| 13 | ], |
| 14 | # nrouter_cache=False, # Uncomment to skip cache |
| 15 | ) |
| 16 | print(response.choices[0].message.content) |
| 17 |
|
| 18 | # Auto-captured metadata |
| 19 | meta = client.last_response |
| 20 | if meta: |
| 21 | print(f"Cost: ${meta.cost}") |
| 22 | print(f"Guardrails: {meta.guardrails}") # x-nr-guardrails: pass, redacted, ... |
| 23 |
|
| 24 | except nRouterGuardrailBlockedError as e: |
| 25 | print(f"Blocked by guardrail: {e}") |
What every app inherits.
Every application built on nRouter gets the same pricing and request-path controls.
Zero platform markup.
Pay as you go is the only plan sold. Usage settles at the served model’s exact list price.
0%
Refusals cost nothing.
Key access, rate limits, budgets, and pre-call guardrails run before a provider is called.
$0
One key for every model.
Switch models by changing one string. nRouter manages the provider keys and bills one balance.
Share your build.
Running an application on nRouter? Tell us how it is built.
01 · Run on nRouter
Run your application on nRouter with active virtual keys and configured budgets.02 · Document architecture
Outline the models, fallbacks, guardrails, and budget scopes your workload uses.03 · Send it to us
Email your architecture notes to the nRouter team.support@nrouter.ai