Unified AI API Gateway
One entry point to international and China-based models. Four-tier quotas — company → department → user → key — are enforced up front, and every call is metered, attributable and auditable.

When to call us
The usual picture: every business line has its own account, key and invoice, and nobody can state total AI spend.
The GSS unified AI API gateway collapses those entry points into one. Applications connect only to the gateway, which handles authentication, quota checks, smart routing, metering and audit; on the model side it connects leading international and China-based vendors.
- Support, marketing, R&D and operations each open their own AI accounts: N platforms, N invoices, and nobody can say what the company spends on AI
- Overspend is found at month end; budgets cannot be enforced in advance or attributed by department or application
- API keys sit with individuals and are not recovered when staff leave or projects end, leaving leakage and runaway-billing risk
- Business code is tied to a single vendor, so price rises, rate limits and outages have to be absorbed
- International models need a compliant, controlled path, while China-based models should be callable through the same interface
One entry point to supported international and China-based models
Enterprise apps connect only to the gateway. International models are connected under your compliance requirements and contract terms; lines, nodes and availability are as agreed in the contract.
YOUR SIDE
- Agent applicationsSupport · marketing · ticketing
- Enterprise appsCRM · OA · ERP · knowledge base
- New workloadsPlug in any time
GSS unified AI API gateway
- Authentication
- Quota check
- Smart routing
- Metering
- Audit trail
MODEL SIDE
- International modelsOpenAI · Anthropic · Google · xAI
- China-based modelsDeepSeek · Zhipu GLM · MiniMax · Moonshot
- Access arrangementsPer compliance requirements and contract
Five common problems, solved
| Problem | Today | What the gateway does |
|---|---|---|
| No single view of spend | N platforms, N invoices | One console aggregates all spend, with CSV export for reconciliation |
| Spend can't be attributed | Nobody knows who spent what | Attributed by department / application / model; every call has an owner |
| Overspend goes unchecked | Found out at month end | Quota up front, real-time checks, hard stop on overspend |
| Keys scattered | Not recovered when staff leave | Mandatory department ownership, one-click revoke, quota released automatically |
| Locked to one platform | Absorb price rises and outages | Multiple vendors natively; switch under pre-set rules |
Four tiers — you decide how the total is spent
Company total → department → user → each API key. Hard rule: allocated at each tier ≤ held at that tier; excess is rejected and overspend stops the call.
Illustrative quota allocation (tokens)
Allocated ≤ held · excess rejected · overspend stops
Five changes the gateway brings
Cost under control
Decide spend in advance. Quota up front, overspend stops.
Transparent spend
See where it goes, attributed by department and application.
Accountable
Know who spends: every call has an owner and a full trail.
Service governance
Routing and switching follow pre-set rules; we do not promise uninterrupted service in all circumstances.
Ready on day one
Protocol-level compatible; no code changes, and compatible apps migrate quickly.
One console for every AI cost
The dashboard shows this month's spend, call counts, token usage (input and output separately) and remaining quota in real time; department detail can be filtered by model and period and exported for reconciliation.
| Module | Capability |
|---|---|
| Billing / usage | Multi-dimension aggregation · CSV export · over-quota alerts |
| Contract orders | Quota issued per contract; one-click issue to the enterprise account once confirmed |
| Model pricing | Leading vendors' models · transparent prices |
| Account management | Tiered company / department / user management |
| Departments | Org-chart tiers · quota ownership |
| API key wizard | Step-by-step creation · mandatory department ownership |
Every key can be issued — and taken back
Create
Name → department → authorised vendors → quota → validity, step by step.
Use
Mandatory department ownership; usable models limited by vendor.
Rotate
External keys suggested to expire in 90 days, with a reminder 7 days before.
Revoke
One click when staff leave or a project ends; quota is released automatically.
Audit
Creation, use and change are logged end to end and traceable.
Why the GSS unified AI API gateway
| Capability | Vendor console | Generic API gateway | GSS gateway |
|---|---|---|---|
| Multiple vendors | ✕ Not supported | ○ Partial | ● Native |
| Tiered quotas | ○ Single-vendor cap | △ Build yourself | ● Four tiers · up-front control |
| Cost attribution | ○ Single-vendor bill | △ Build yourself | ● Department / project / key |
| Quota from contract | ✕ Not supported | △ Build yourself | ● Issued per contract |
| Call audit | ○ Single-vendor logs | ○ HTTP layer | ● End to end · multi-dimension filter |
| Developer compatibility | ● Native SDK | △ Needs adaptation | ● Protocol-level compatible |
● Native ○ Partial △ Build or adapt yourself ✕ Not supported
Specifications
- Model access
- OpenAI / Anthropic / Google / xAI; DeepSeek / Zhipu GLM / MiniMax / Moonshot. Protocol-level compatible — no code changes on the business side
- Gateway functions
- Unified authentication, quota checks, smart routing, precise metering (input and output tokens counted separately) and end-to-end audit trail
- Four-tier quotas
- Company total → department → user → each API key. Allocated at each tier ≤ held at that tier; excess is rejected, overspend stops the call
- Admin console
- Billing and usage aggregation with CSV export, one-click quota issue from contract orders, model pricing, tiered account and department management, API key wizard
- Key lifecycle
- Create → use → rotate (90 days suggested, reminder 7 days ahead) → revoke → audit
- Availability & compliance
- Routing follows pre-set rules; network access and data transfers follow your compliance requirements. Lines, nodes and availability are as agreed in the contract
- Delivery model
- GSS-hosted (SaaS) or private deployment in your data room or dedicated cloud; quotas are issued per contract, with contracting, settlement and invoicing handled by GSS
How delivery works
Connect
Point existing agents and enterprise apps (support, marketing, ticketing, CRM / OA / ERP / knowledge base) at the gateway. No business interruption; usually done the same day.
Allocate
Build the four-tier quota to match your org chart, enforce department ownership and limit the models each key can use by vendor.
Track
Spend, call counts, token usage and remaining quota are visible in real time, filterable by department, application, model and period, and exportable.
Govern
Over-quota alerts and automatic blocking, periodic key rotation and one-click revocation, monthly cost attribution reports.
In production
Frequently asked questions
Didn't find your answer? Ask us
What does the gateway add over using vendor consoles directly?
Do we need to change business code?
How are quotas controlled?
What happens to keys when staff leave or a project ends?
How do you balance availability and compliance for international models?
How is it paid for?
Is this a GSS-operated product?
Tell us what you need
Send us site addresses and requirements. A solutions architect will reply within one business day.
- 1Reply within 1 business dayFrom a solutions architect, not a ticket queue.
- 2Quote in 1–3 business daysRoute options, alternatives and per-segment costs.
- 3Contract the way you preferHong Kong entity in USD, or mainland entity in RMB with VAT invoice.
Or email sales@goldsunshine.net