From intent to execution

AI agent Codex keeps complex work moving

Stable Codex access for users in Mainland China without network hurdles or overseas payment friction. Start with Codex App so the agent can break down requirements, edit across files, run checks, debug failures, and summarize delivery in one conversation.

Get startedRead docs
Codex
Current threadComplex delivery
Plan readyVerifying
You

Help me move a complex project from idea to a shippable version.

CodexAgent is working

I will build a task plan, read context, edit the right files, run checks, and summarize the shippable result.

Agent execution plan4 stages · 8 context files · test verification
  1. Scope the requestClarify goals, constraints, dependencies, and acceptance checks.
  2. Plan the implementationBreak work into steps while mapping risk and impact.
  3. Edit across filesUpdate pages, styling, copy, and interaction states.
  4. Verify and hand offRun checks, fix failures, and summarize delivery.
Ask Codex to continue with the next step...

why get-codex

Why choose Get-Codex

Not a bigger model list. Not a generic relay. We focus on real usage and practical delivery, making Codex easier to access, faster to use, and steadier to rely on.

  1. No complicated network setup

    Designed around real Mainland China access conditions, reducing the friction of network prerequisites.

  2. Extreme speed

    Asia-Pacific tri-network optimized routing, 200 Mbps bandwidth, as low as 60 ms latency.

  3. No overseas payment needed

    Domestic payment options with invoice support.

  4. Data security

    All traffic encrypted via HTTPS. Data plane and control plane are separated — user data is never retained or analyzed. Critical control plane data is encrypted with AES.

reliability

Reliability and safety

Verifiable at every layer, from network to data.

  1. Single Codex link

    All resources dedicated to the Codex access path, eliminating stability overhead from multi-model routing.

  2. Transparent and auditable billing

    Real-time quota visibility with full usage and billing records. Self-service plan upgrades and downgrades.

  3. Zero data retention

    End-to-end HTTPS encrypted forwarding. No logging, storage, or analysis of any request or response content.

Access

Supported access ways

Start with Codex App for a more direct agent interface, then use Codex CLI only when terminal automation fits better.

  1. Codex App

    Codex App

    The recommended entry point. Sign in and manage agent tasks, project context, and execution progress in the desktop interface.

  2. Codex CLI

    Codex CLI

    An optional entry point for scripting, terminal automation, or preserving an existing command-line workflow.

FAQ

The questions people ask most often when getting started with Codex, all in one place.

Does it support Codex CLI?

Yes. You must set OPENAI_BASE_URL to https://api.get-apikey.com/v1 (including /v1). Add your issued API key, and the Codex CLI runs as usual — no change to your existing workflow. The same key also works against the Responses API, and both paths share the same quota.

Do I need extra network workarounds first?

No. Traffic runs over tri-network optimized direct lines with 200 Mbps bandwidth and 60 ms floor latency, so Codex CLI and the Responses API stay reachable from Mainland China without a VPN or proxy.

How does billing work?

Each plan ships with a Codex call quota. Live usage details and debit records are visible from the dashboard, and you can upgrade or downgrade on demand. Domestic payment methods are supported and invoices are available — see the pricing page for the tier breakdown.

Is my data safe?

All traffic is forwarded over HTTPS; request and response bodies are never logged, stored, or analyzed. The data plane and management plane are fully isolated, sensitive fields on the management plane are stored with AES encryption, and nothing you send is ever used for model training.

Why focus on Codex?

We invest the full engineering surface into a single Codex path — optimized routing, billing, rate limits, and data-plane isolation — instead of spreading across a generic multi-model catalog. One path taken to its limit is more valuable for high-frequency Codex usage than a long list of sampler endpoints.

Start using Get-Codex

Create an account, follow the setup guide, and start using Codex. If you are still evaluating, read the docs or the integration guide.

Get startedRead docs