GateLLM
Telegram

Private pilot

Fast gateway
for Codex.

One managed endpoint for Codex and OpenAI Docs MCP. You continue to sign in with your own OpenAI/ChatGPT account.

Get access How to connect
  • Streaming responses
  • Your own account
  • Codex + Docs MCP

Setup

From Telegram to Codex
in four steps.

Save your current configuration first. Then add GateLLM as a custom provider.

  1. 01

    Sign in to Codex

    Make sure Codex is already authenticated with your own OpenAI/ChatGPT account.

  2. 02

    Get your token in Telegram

    Open the official bot and receive your personal GateLLM token.

    Open @GateLLMBot ↗ REPLACE_WITH_TOKEN_FROM_TELEGRAM
  3. 03

    Back up and update your configuration

    Copy ~/.codex/config.toml somewhere safe, then add the configuration shown here. Replace the token marker in both places.

  4. 04

    Restart Codex

    Open a new session. GateLLM is ready to receive requests through the configured provider.

~/.codex/config.toml 
model = "gpt-5.6-sol"
model_reasoning_effort = "high"
service_tier = "default"
model_provider = "gatellm"

[model_providers.gatellm]
name = "GateLLM Codex gateway"
base_url = "https://gatellm.online"
wire_api = "responses"
requires_openai_auth = true
http_headers = { "X-Token" = "REPLACE_WITH_TOKEN_FROM_TELEGRAM" }

[mcp_servers.openaiDeveloperDocs]
url = "https://gatellm.online/openai-docs-mcp"
http_headers = { "X-Token" = "REPLACE_WITH_TOKEN_FROM_TELEGRAM" }
startup_timeout_sec = 20
tool_timeout_sec = 120

Before editing: save a copy of your existing file. Changing providers does not delete history, but older sessions may not continue under the new provider exactly as they did before.

Store the token more securely

Instead of both http_headers lines, you can use env_http_headers = { "X-Token" = "GATELLM_TOKEN" } and supply the token through a local environment variable.

Clear boundaries

Transparent about traffic.

01

Your sign-in credentials

Never send your OpenAI password, session token, authorization header, or cookies to the Telegram bot.

02

Gateway traffic

GateLLM terminates TLS and can technically observe passing traffic. It is designed not to log prompts, responses, or upstream credentials.

03

Landing-page analytics

Yandex Metrika runs on this public page. Its analytics code is not part of Codex requests to the API gateway.

Private pilot

Ready to connect Codex?

Start with the official GateLLM Telegram bot.

Open @GateLLMBot
QR code linking to the official GateLLM Telegram bot