For Codex and Claude Code Developers
Rate Limit hit again?
Use HaoAI, keep Codex running
3-second setup, keep Vibe Coding
# Just add two lines
export OPENAI_BASE_URL=https://api.hao.ai/v1
export OPENAI_API_KEY=HAOAI_API_KEY
# Keep using Codex through HaoAIWorks with coding agents
Why HaoAI?
Three hard metrics, benchmarked against official API
No Rate Limits
Unlimited Throughput
No RPM limits. No TPM limits. No queuing during peak hours. Long context? No problem. We protect your vibe coding flow.
GPT-5.5 / Claude Opus 4.8 and current flagship models supported
Zero Migration Cost
Native Protocol Compat
Full Anthropic Messages API and OpenAI-compatible protocol support. Codex, Claude Code, and Cline plug in without adapters.
Supports OpenAI / Anthropic / Google protocols
Enterprise Reliable
99.9% SLA
Multi-node redundancy with automatic failover. Globally accelerated routes keep latency around ~300ms for critical demos.
Monthly uptime 99.9%+, P99 latency <200ms
30-second setup
Codex
1
Get HaoAI API Key2
Configure the HaoAI provider3
Enable WebSocket# ~/.codex/config.toml
model_provider = "haoai"
[model_providers.haoai]
base_url = "https://api.hao.ai/v1"
wire_api = "responses"
supports_websockets = trueClaude Code
1
Get HaoAI API Key2
Set env variable3
Keep vibe coding# ~/.zshrc or ~/.bashrc
export ANTHROPIC_BASE_URL=https://api.hao.ai/anthropic
export ANTHROPIC_AUTH_TOKEN=HAOAI_API_KEY~300ms
Global acceleration
99.9%
Monthly uptime
Flagship
GPT / Claude
3 Protocols
OpenAI / Anthropic / Google