Documentation
ChatGPT API - what it is and how to start using the OpenAI API in 2026
Last updated:
What is the ChatGPT API?
The ChatGPT API is a third-party, OpenAI-compatible gateway that gives you access to the latest OpenAI models through a single API key. It speaks the official OpenAI request and response format, so any existing OpenAI integration keeps working after you change one setting: the base URL.
ChatGPT API is an independent provider and is not affiliated with or endorsed by OpenAI.
What does ChatGPT API provide?
ChatGPT API gives access to every OpenAI model through one key. It uses the official OpenAI format, so you can connect it to any SDK, code editor, agent, or automation tool - only the base URL changes.
| Capability | Description |
|---|---|
| OpenAI-compatible | POST /v1/chat/completions, POST /v1/responses, GET /v1/models with SSE streaming |
| OpenAI models | GPT-5.6-sol, GPT-5.6-terra, GPT-5.6-luna, GPT-5.5, GPT-5.4 and GPT-5.4-mini |
| Integrations | Cursor, GitHub Copilot, opencode, Codex CLI, VS Code, SDK, cURL |
| Payments | 50% cheaper than official OpenAI rates, pay with crypto |
ChatGPT API vs the official OpenAI API
You keep the same code, SDKs and request format. The differences are commercial:
| Official OpenAI | ChatGPT API | |
|---|---|---|
| Pricing | List per-token rates | 50% below official, per token |
| Billing | Subscription or card required | Pay-as-you-go, credits never expire |
| Payment | Card | Crypto - BTC, ETH, USDT, USDC |
| Setup | api.openai.com | Change only the base URL |
| Models | OpenAI only | Same OpenAI models, one key |
How to get a ChatGPT API key
- Register a free account and open your dashboard.
- Click Generate key to create your ChatGPT API key.
- Set it as your API key and point the base URL at our endpoint - no approval queue, no waiting.
Recommended model IDs
ChatGPT API exposes the full OpenAI model lineup through one key and the official OpenAI format, so you choose a model purely by the speed, cost and quality your task needs. The new GPT-5.6 family leads: gpt-5.6-sol is the flagship for the hardest reasoning and coding, gpt-5.6-terra is the balanced everyday choice, and gpt-5.6-luna is the fastest and cheapest. The previous-generation gpt-5.5, gpt-5.4 and gpt-5.4-mini remain available. Pass any of these IDs in the `model` field; switching between them is a one-line change.
gpt-5.6-solnew flagship, best for complex reasoning and codinggpt-5.6-terrabalanced quality and price for everyday workgpt-5.6-lunafastest and cheapest of the new familygpt-5.5previous flagship, strong all-round balancegpt-5.4solid price/quality baselinegpt-5.4-minifast and cheap workhorseCodex API: models for vibe coding
For coding, refactoring and agentic workflows, use the new gpt-5.6-sol as the flagship or gpt-5.5 as a solid all-rounder. gpt-5.6-luna gives a great speed-to-cost balance for high-volume completions. They all work through the same OpenAI format, so you can plug them into Codex CLI, Cursor, opencode, GitHub Copilot or your own agent by changing only the base URL and the model ID.
How much does the ChatGPT API cost?
Pricing is pay-as-you-go and mirrors the official OpenAI per-token rates minus 50%, on both input and output tokens. There is no subscription and no minimum spend - you buy credits once, they are consumed only when a request runs, and they never expire.
Payment and access without official billing
Because top-ups are paid with cryptocurrency, you can use the OpenAI API even where official card billing isn't available. Buy credits once, use them across every model, and track token-level spend in your dashboard.
Frequently asked questions
How much does the ChatGPT API cost?
The ChatGPT API is pay-as-you-go and priced 50% below the official per-token list price. There is no subscription and no minimum spend - you buy credits once and they are only consumed when a request actually runs. Every request is metered at token level in your dashboard, so a small project can cost just a few cents. Pricing mirrors the official OpenAI per-token rates exactly, minus 50%, on both input and output tokens, and cached input is billed at the same discounted ratio. Because billing is usage-based rather than seat- or tier-based, your costs fall to near zero while your app is idle and scale up smoothly as traffic grows. You can see the full per-model breakdown on the pricing page and track spend in real time from your dashboard.
Is the ChatGPT API free?
Not entirely, but you can try it without paying first. New accounts get a $0.25 test balance right after sign-up - no card, no subscription - so you can generate a key, send real requests to any model, and see actual token-level billing in your dashboard before spending anything. Once that test balance runs out, the API stops working until you top up your balance with crypto (BTC, ETH, USDT, USDC); there's no automatic charge and no recurring fee. Every request after that is billed per token at 50% below the official OpenAI rate. Credits you buy never expire and there is no minimum spend, so a small side project can run for months on a few dollars.
Which models are available?
You get drop-in access to the latest OpenAI lineup - the new GPT-5.6 family (sol, terra, luna) plus GPT-5.5, GPT-5.4 and GPT-5.4-mini - through the official OpenAI format. Because we are API-format compatible, the model names you already use keep working. You pass the model ID in the standard `model` field of your request, exactly as with the official API, so switching between models is a one-line change. The full list of model IDs, with a short note on when to use each, is on the Models page in the docs.
Which model is best for coding?
For the hardest coding, refactoring and agentic tasks, gpt-5.6-sol is the strongest choice - the new flagship of the 5.6 family. gpt-5.5 remains a solid all-rounder. For faster, cheaper day-to-day coding and high-volume completions, gpt-5.6-luna and gpt-5.4-mini give the best balance of speed and cost.
How do I get a ChatGPT API key?
Register for a free account, open your dashboard, and click “Generate key”. Your ChatGPT API key works immediately - set it as your API key and point the base URL at our endpoint. No approval queue and no waiting.
Which base URL do I use?
Set the base URL in your SDK or tool to https://api.llm-gate.tech/v1.