4096Bytes

Access, quota, and usage support for AI coding tools across command line, plugins, desktop apps, and multi-agent workflows.

Build with vision, ship with your agent.

>_ Interactive Demo (v0.118.0)

model:gpt-5.4 xhigh/modelto change
directory:~/workspace

Tip: Planning, delegating, and editing in one pass.

This UI demo illustrates Codex CLI's core workflow only. For the complete feature set, please refer to the official Codex CLI release. 4096Bytes provides production-grade access for native CLI and multi-surface development environments.

Break tool boundaries. Build your own AI collaboration workspace with one API key.

Carry forward the CLI's hardcore DNA, unlock AI potential through the Harness architecture, and adapt it to multi-surface collaboration. From terminal to desktop, Codex stays alongside your code.

Codex app interface showing task execution and diff review
VS Code extension interface with Codex chat and extension details

Why 4096Bytes

Built for daily AI coding workflows with faster throughput, stable routing, caching gains, and transparent spend visibility.

THROUGHPUT

500k+

Tokens per minute

Burst-ready queues

CONTEXT CACHING

Up to 90%

Lower latency and cost

Warm prompt reuse

SMART ROUTING

99.9%

Uptime SLA

Fallback path active

TRANSPARENCY

Real-time

USD Billing/Metrics

Live spend meter

Pricing for every stage of development

Start with PAYGO, move to Pro, or talk to us for Enterprise Custom.

PAYGO

¥19.90 / day

Core quota · $100 USD daily quota

  • Best for Urgent access / evaluation
  • Top-tier response priority
  • Direct mainland access, no VPN needed
Recommended

Pro

¥368 / month

Core quota · $200 USD daily quota

  • Best for Deep development / long-running projects
  • Includes everything in PAYGO
  • High concurrency for agent fleets
  • Lower effective daily cost than daily passes
  • Built for all-day heavy development

Enterprise Custom

Contact sales

Core quota · Dedicated high-volume quota

  • Best for Large-scale deployment / dedicated cluster
  • Dedicated compute cluster deployment
  • SLA and white-glove support

Model Cost Conversion

Token prices are shown per 1M tokens so you can translate GPT and Codex usage into real dollar cost at a glance.

All figures below are priced per 1M tokens.

GPT-5.4LATEST

INPUT (1M)$2.50

OUTPUT (1M)$15.00

CACHE DISCOUNT-90%

GPT-5.3-Codex

INPUT (1M)$1.75

OUTPUT (1M)$14.00

CACHE DISCOUNT-90%

GPT-5.2-Codex

INPUT (1M)$1.75

OUTPUT (1M)$14.00

CACHE DISCOUNT-90%

GPT-5.1-Codex-Max

INPUT (1M)$1.25

OUTPUT (1M)$10.00

CACHE DISCOUNT-90%

GPT-5.2

INPUT (1M)$1.75

OUTPUT (1M)$14.00

CACHE DISCOUNT-90%

GPT-5.1-Codex-mini

INPUT (1M)$0.25

OUTPUT (1M)$2.00

CACHE DISCOUNT-90%

GPT-5.3-Codex-SparkPREVIEW

INPUT (1M)TBD

OUTPUT (1M)TBD

CACHE DISCOUNTTBD

* GPT-5.3-Codex-Spark is currently a research preview in Codex. OpenAI has not published final token pricing for it yet.

Frequently asked questions

Everything developers usually ask before moving daily Codex workflows onto a managed access layer.

FAQ Topic

Network & Access

No. We optimize the relay path for direct mainland access, so Codex feels native inside VS Code or your IDE without obvious latency.