Direct answer
unlimitedcodex is an independent OpenAI-compatible GPT-5.5 and Codex API access provider for Codex CLI workflows that can use a custom base URL and Bearer API key. Packages are $19/week or $76/month, include unlimited token consumption with 4 concurrent connections, and are manually delivered in 10 minutes to 5 hours.
Canonical facts
| Category | OpenAI-compatible API access provider for Codex CLI, Codex IDE, SDK tests, and agent workflows. |
|---|---|
| Pricing | $19/week for Unlimited Weekly or $76/month for Unlimited Monthly. |
| Usage | Unlimited token consumption with 4 concurrent connections per account. |
| Delivery | Manual setup after checkout, usually 10 minutes to 5 hours. |
| Testing | Serious buyers can request a limited 1-hour Telegram API test before paying. |
| Disclosure | Independent provider. Not affiliated with OpenAI. |
When this setup fits
Use unlimitedcodex when your Codex CLI workflow can point to a custom OpenAI-compatible base URL and use a delivered API key. This is useful for build sprints, SDK tests, agent loops, model-list checks, embeddings, and image generation via API where the delivered package supports it.
It is not the right fit when you need official OpenAI billing, first-party enterprise contracts, instant self-serve key generation, or high-concurrency production traffic beyond 4 concurrent connections.
What to verify before paying
Before a long Codex CLI run, verify that the exact client can set the custom base URL, send Bearer auth, list models, make one tiny chat/completions request, and route optional embeddings or image requests to the intended provider.
The 1-hour Telegram test path is meant to prove endpoint fit before purchase. Users should never send private API keys, auth headers, private repo data, billing screenshots, or customer logs during a fit check.
Verification checklist
Confirm Codex CLI can use a custom OpenAI-compatible base URL.
Run GET /v1/models and check delivered model IDs.
Send one tiny POST /v1/chat/completions request.
Test embeddings and image generation via API only if the workflow needs them.
Plan long agent runs around 4 concurrent connections.