For Codex, Claude Code, agents, and OpenAI-compatible clients
Sell reliable AI access, not another API key list.
CodexManager Relay turns multiple model vendors, aggregate APIs, usage logs, billing rules, and failover policies into one private gateway your customers can subscribe to.
- 23ms
- local routing overhead target
- 12+
- provider and aggregate API profiles
- 99.9%
- policy-level availability objective
Codex Desktop
Relay Gateway
gpt-5.5 -> azure2
OpenAI Compatible
GLM Aggregate
Azure fallback
Built for operators who need to package AI routing as a paid product: one gateway, clear customer tiers, predictable failover, and evidence-grade logs.
Sell subscriptionsSeats, request volume, and private deployment licenses.
Protect uptimeHealth checks, cooldowns, retries, and fast fallback.
Control costTrack spend by model, vendor, client, and team.
Who it is for
A commercial control layer for teams already using AI tools every day.
The page is designed for selling a product, so it explains outcomes before implementation details.
AI development teams
Give Codex, Claude Code, Cursor-style clients, scripts, and SDKs one controlled endpoint with per-team usage visibility.
Model resellers
Package aggregate API capacity into tiered plans with routing rules, usage evidence, and support diagnostics.
Private deployment buyers
Keep keys and traffic inside the customer environment while still receiving updates, support, and license control.
Commercial feature set
Everything needed to turn a relay into a subscription product.
Each capability is written for real operations: easy to sell, deploy, monitor, and support.
Unified client endpoint
Expose one base URL for Codex Desktop, Claude Code, OpenAI-compatible SDKs, and automation agents.
Model routing matrix
Route by model, vendor priority, protocol capability, health state, and customer policy instead of manual config edits.
Fast failover engine
Cool down unhealthy upstreams, retry transport failures, and fallback before the customer experience stalls.
Billing and cost panels
Estimate cost by token, model, upstream, team, and plan so renewals and internal chargeback are defensible.
Request logs and audit trail
Capture client signals, route attempts, status codes, error classes, and upstream decisions for support workflows.
Private deployment license
Deploy locally, on team servers, or inside a customer network while keeping upstream keys under customer control.
Operator cockpit
The selling point is not only routing. It is operational confidence.
Provider health board
Show active, cooling, degraded, and disabled vendors with probe history and last error reason.
Customer-facing usage exports
Export request volume, token usage, cache usage, and model mix for customer reports or invoices.
Routing policy presets
Create default presets for low-cost, high-reliability, image generation, reasoning-heavy, and fallback-only workloads.
Support-ready diagnostics
Open a failed request and see the client, model, selected route, upstream response, retry path, and final outcome.
How it works
One chain from purchase to production operations.
The implementation can start small, then grow into private deployment, customer accounts, and managed support.
01Choose a license
Pick Starter, Pro, or Business by team size, request volume, and deployment model.
02Import providers
Add aggregate APIs, OpenAI-compatible endpoints, native vendors, model aliases, and prices.
03Point clients to Relay
Clients only need the Relay base URL and key; routing and fallback are handled by the gateway.
04Operate with evidence
Use health, billing, logs, and exports to support customers and refine route strategy.
Subscription plans
Buy the gateway layer without locking yourself to one model vendor.
These are marketing-page numbers for presentation. Replace them with the final commercial policy when ready.
Starter
Solo builders and early trials
$19/mo
- 3 upstream providers
- 300K requests per month
- Basic request logs
- Community update channel
Recommended
Pro
High-frequency AI teams
$79/mo
- Unlimited providers and model routes
- 3M requests per month
- Failover and health probing
- Billing panels and exports
Business
Private deployment and customer licensing
$249/mo
- Private deployment license
- Team seats and audit logs
- Dedicated deployment support
- SLA and custom adapters
ROI calculator
Estimate the time and incident cost your team can avoid.
Move seats and monthly request volume. The page estimates a plan, subscription cost, and saved operator time.
Recommended planPro
Estimated monthly fee$79
Estimated monthly savings$736
Why not DIY?
A gateway product should remove repeated operational work.
Need
Manual API sharing
CodexManager Relay
Switch broken upstreamsUsers edit config after failures.Gateway cools down and falls back.
Explain customer billsSpreadsheet reconstruction.Usage, token, and vendor evidence.
Support failed requestsAsk for screenshots and guess.Trace route attempts and upstream errors.
Add new model vendorsEach user updates clients.Operators update policy once.
Trust layer
Security boundaries that make commercial deployment easier to approve.
Upstream API keys are stored in the customer environment, not in a hosted marketing system.
Client, model, vendor, status, error class, and fallback decisions remain visible for support.
Use OpenAI-compatible endpoints, aggregate APIs, private model services, or native vendor profiles.
Business deployments can support offline licensing, fixed-version maintenance, and intranet delivery.
What customers receive
A package that looks like a product, not a one-off script.
Gateway software
Deployable relay service with routing, health, logs, billing, and configurable upstream profiles.
Operator handbook
Setup, route policy, failure triage, upgrade, backup, and customer onboarding guidance.
Support SLA option
Defined response windows, incident evidence requirements, and custom adapter development path.
Get started
Request a trial or private deployment quote.
This static form gives immediate front-end feedback. In production it can be wired to CRM, email, payment, or license provisioning.
FAQ
Questions buyers ask before subscribing.
Is this a replacement for CodexManager?
Commercially, it is positioned as a productized relay layer around CodexManager-style routing: subscription packaging, billing, audit logs, and private deployment support.
Do customers need to change every client?
Usually no. Customers point clients at the Relay base URL and use an authorized key. Model selection, vendor routing, and fallback policy are handled by the gateway.
Does the subscription include upstream model cost?
By default, no. The subscription covers software licensing and operational capability. Upstream model usage is billed through the customer's vendor or aggregate API account.
Can it run inside a customer network?
Business plans are designed for private deployment and can be extended with offline licensing, fixed-version maintenance, internal audit, and custom provider adapters.
What makes it suitable for resale?
It provides plan language, usage evidence, route policy, diagnostics, and support boundaries, so operators can sell a managed gateway instead of unmanaged API credentials.