Welcome
Your supply-first control plane — suppliers, keys, routing and spend in one place.
My plan
Loading your limit…
One API. Your next idea.
Make your first call in three steps — then hand BharatRouter to your AI agent or IDE.
- 01 Create a gateway key
- 02 Connect your app or AI IDE
- 03 Make your first call
https://api.bharatrouter.com/v1Using Claude Code or Cursor? Copy the agent setup and add your br- key to get connected.
Make it yours
Read the docsAPI keys
Your gateway to every model. Create a br- key, connect your app, and start building.
Gateway keys
Keys created by people in your workspace, with individual limits and expiry.
Loading…
Application keys ORG
Shared credentials for services, pipelines, and agents.
Service credentials owned by the organisation — not tied to any person. They survive member changes and any admin can revoke them. Use these for CI/CD, backend services, and agents.
Settlement & usage
Per-application-key consumption for intra-company chargeback — requests, tokens and notional ₹ usage-value (a statement, not a tax invoice), broken down by model.
Loading…
Usage
Requests, tokens, and spend — a clear view of how your workspace uses AI.
Loading…
Logs
Explore request metadata, costs, and trace IDs. Inspect stored payloads when content logging is enabled for your organisation.
Loading…
Traces
Connect requests across traces, sessions, and agents. Inspect latency waterfalls and verify where your traffic was served.
Loading…
Billing
Prepaid INR credits, add-credits, and your GST tax invoices.
Loading…
Workspaces
Operational environments inside the org — scope keys, budgets and usage. Keys with no workspace live in the org default.
Loading…
Sangams
Combine models into a single callable ensemble. Use a built-in Sangam, or create a chat or image ensemble for your organisation.
Loading…
Agents
Manage revocable identities for your automations. Use short-lived tokens, attribute usage per agent, and set monthly budgets.
Loading…
Account settings
Your profile and this organisation’s name.
Loading…
Admin
Founder-only — grant goodwill/promo credit, adjust organization monthly budgets, manage provider permissions, and review the audit log. Every change is audited.
Loading…
GPU Leads
Founder-only — orgs whose inference footprint signals a GPU-cluster / dedicated-capacity conversation. Scored from usage volume, growth, India/Krutrim share and consistency. Read-only demand-sensing; no money or entitlements change here.
Loading…
Suppliers & Catalog
Which providers this org can route to, and whose key. Platform models need no setup; bring your own key for third-party providers.
Loading…
Routing & Fallback
How requests pick a model and fail over. Smart defaults unless you pin a chain per model.
Loading…
Playground
Run a prompt across one or more models side by side — with real latency, tokens and ₹ cost. Goes through the same routing, guardrails and structured-output pipeline as production.
Loading…
Cost
Understand spend in five seconds — then drill into any cost question.
Overview ·
💬 Ask your usage metadata-only
Natural-language FinOps over your traces — which agent burned the most budget, what share was served in India, the priciest sessions on a non-India provider. Never your prompts or responses.
Cost over time
Daily ₹ spend across the window — the headline the live dashboard lacks.
BYOK-real is an estimate at catalog list price (or your custom price), not the provider's invoice.
Cost drivers
Top spenders by dimension — including the agent / workspace / key attribution incumbents lack.
Provider spend split
Provider mix at a glance — Krutrim-first.
Budget burn-down ·
Cumulative spend vs the linear-pace ideal and the month's cap — are we on track this month?
Budget caps & reconciled status
Included · consumed · remaining · over — org-wide and per scope.
Cost-efficiency trend
₹ / request and ₹ / Mtok over time — is a model getting pricier per token?
Residency cost split NL-backed
India vs Global vs Unknown — the India-first split no incumbent answers.
Model pricing estimate-only
Set a ₹/Mtok price for your BYOK / self-hosted models so BYOK-real renders a real figure instead of ₹0. This shapes the estimate only — it never bills you, debits your wallet, or trips a budget cap. Owner/admin only.
Needs a price — recently-used BYOK models the catalog leaves at ₹0 (zero-token requests also read ₹0)
Custom prices
All ₹ figures are realized spend (usage_events.cost_uinr), IST day boundaries. Org-cost panels need the cost:org:view Finance capability. BYOK-real is an estimate at catalog list price (or your custom price), not the provider's invoice.
Adoption
Who's actually using BharatRouter — active developers, the adoption tail to chase, and what each dev leans on.
Overview ·
AI adoption over time
Distinct active developers per IST day (left) and adoption % of members (right) — the trend the live dashboard lacks.
Daily active developers
Distinct developers with BR-routed traffic each IST day.
Model mix
Which models the org leans on. By requests — no Finance access needed.
Adoption by team / workspace
Active developers per workspace. Membership is key-based (api_keys.workspace_id), not HR org.
No-data members
Members with zero BR-routed traffic in this window — the adoption tail to chase.
Adoption = which org members have BR-routed traffic (IST day boundaries). BR is a gateway — no code-output or shadow-AI metrics. Spend tiles need the cost:org:view Finance capability; the whole page needs usage:org:view (own card otherwise).
Experiments
Run a golden dataset across prompt variants — see which wins on pass-rate, cost and latency, then promote it.
Pass-rate
Share of rows each variant passed. Winner in accent.
Mean cost
Notional ₹ per row (usage-value, not a bill).
Latency
p50 / p95 response time (ms).
Per-variant
One row per variant. Promote repoints the winning prompt's prod tag. Owner/admin only.
| Variant | Prompt / model | Passed | Pass-rate | Mean ₹ | Total ₹ | p50 ms | p95 ms | Errors | |
|---|---|---|---|---|---|---|---|---|---|
| No run selected. | |||||||||
Notional ₹ is usage-value, not a charge. Experiment traffic rides the loopback self-call so routing, prompt-render and residency pinning apply exactly as production.
Spend & Budgets
Track usage, review monthly caps and keep your workspace spending in view.
Loading…
Domains & SSO
Independently DNS-verified domains attached to this org — auto-join & SSO.
Loading…
Key Vault
Manage your provider credentials, access and monthly spend caps in one place. Your keys stay masked when shared for use.
Owners and admins manage provider credentials.
Share access across your organisation, a team or a workspace. The secret stays sealed.
Provider keys
Find a provider and manage who can use its key.
Loading keys…
Leaderboards
Adoption & efficiency — ranked on tokens, then requests. Never on raw spend.
Loading…
Insights
Governance analytics — agent (NHI) usage, shadow-AI adoption, and spend/token/error anomalies vs a trailing baseline.
Loading…
DPO / Compliance
Review processing activity, data residency and retention, then export your Data Protection Impact Assessment (DPIA). Evidence contains metadata, without prompt or response bodies.
Loading…
Members
People in this org, their role, and any per-member monthly budget.
Loading…
Organization
This org’s identity and settings.
Loading…
Audit log
Every control-plane action in this org — who did what, when, from where. Append-only and content-free (never request/response bodies). Filter, page, and export for your ITGC / DPDP evidence.
Audit events
Filter recorded actions, browse pages, or export the matching evidence.
| When | Actor | Action | Target | Detail | IP |
|---|---|---|---|---|---|
| Loading… | |||||
Webhooks
Register HTTPS endpoints to receive signed event notifications — budget, keys & access, billing, and reliability. Each POST carries an X-BR-Signature header; payloads are metadata only (never request/response content). Docs & signature verification →
Loading…
Become a provider
Bring your models to BharatRouter. Set up your provider profile, add an endpoint, and submit your listings for review. Provider guide →
Loading…