How to Use the GPT-5.5 API for Free

GPT-5.5 launched on April 23, 2026, behind a paywall for most users—Plus, Pro, Business, and Enterprise plans in ChatGPT, and paid API tokens for programmatic use. However, there are three current, verifiable ways to access GPT-5.5 for free. If you can work within rate limits and accept temporary access, you can make actual GPT-5.5 calls without adding payment details.

Try Apidog today

This guide covers all free options, maps them to use cases, and shows how to set up a production-ready request collection in Apidog to ensure a smooth transition when you need to move off the free tier.

TL;DR

• Codex CLI on ChatGPT Free or Go — Use the Codex command-line tool for free, temporary GPT-5.5 access. No credit card required.
• OpenAI trial credit for new API accounts — New accounts get a small starter balance, usable for GPT-5.5 calls once the Responses API is available.
• OpenRouter and aggregator free tiers — Some third-party gateways offer free quota for new models shortly after launch.
• All paths have usage caps. For any production workload, switch to paid billing before free access runs out.

Path 1: Codex CLI (the most practical free route)

OpenAI includes Codex with every ChatGPT plan (Free and Go, limited time). Codex exposes GPT-5.5 using ChatGPT sign-in, not an API key. Sign in with a free account, run the CLI, and interact with GPT-5.5 using a 400K-token window.

Install

npm install -g @openai/codex
# or
brew install codex

Verify the installation:

codex --version

Authenticate

First run of codex opens a browser for ChatGPT OAuth. For headless environments:

codex login --device-auth

This uses device code authentication—copy the URL and code to another machine to authorize. No API key is needed.

Pick the model

Within Codex:

/model gpt-5.5

Or start Codex directly with the model flag:

codex --model gpt-5.5

Check your remaining quota with /status. Free and Go plans have tighter limits than paid, but enough for prototyping.

What is available (and what is not)

• You get: Actual GPT-5.5 model, 400K context, file reads, terminal command execution, repo editing within the CLI.
• You do not get: Direct API access. GPT-5.5 is only accessible through Codex when signed in.

For a step-by-step Codex guide, see our free GPT-5.5 with Codex guide.

OpenAI's "limited time" window means Codex access for Free and Go users will end. Make your project flexible by letting the model ID be configurable—swap to a paid plan or API when needed without rewriting code.

Path 2: OpenAI trial credit for new API accounts

New OpenAI developer accounts typically receive a small trial credit. The amount can vary—historically $5 for 90 days, more for .edu emails. When GPT-5.5 API is generally available, you can use this trial balance for real calls.

How to claim trial credit

1. Create a new developer account at platform.openai.com using a fresh email (previous billing history may disqualify you).
2. Verify your phone number. This is required for the trial.
3. Create a project-scoped API key under the trial organization.
4. Check the usage dashboard for your credit amount and expiry date.

GPT-5.5 usage with trial credit

With $5 credit and current pricing ($5/M input, $30/M output tokens), you get about 1M input tokens or ~160K output tokens with GPT-5.5. Enough to prototype, benchmark, or validate workflows—not enough for production.

Save costs during trial

• Use Batch mode: Batch API requests cost 50% less. Great for workflows not requiring instant responses. See OpenAI API pricing.
• Set reasoning.effort to low: Default is low (same as GPT-5.4). Higher values consume more tokens.

Important limitations

• Trial credit is one-time. When used up, calls return 402. You cannot get a second trial on the same payment or phone credentials.

Path 3: Aggregator free tiers

Third-party gateways (OpenRouter, Together, Groq) sometimes offer free quota for new models. These offers are temporary—check availability before relying on them.

Typical setup pattern

1. Sign up and verify your email with the aggregator.
2. Obtain an aggregator API key.
3. Change your SDK’s base URL to point to the aggregator.
4. Use the aggregator’s model alias, e.g., openai/gpt-5.5.

Example with OpenRouter:

from openai import OpenAI

client = OpenAI(
    base_url="https://openrouter.ai/api/v1",
    api_key="sk-or-v1-...",
)

response = client.chat.completions.create(
    model="openai/gpt-5.5",
    messages=[{"role": "user", "content": "Explain the Responses API in two paragraphs."}],
)

print(response.choices[0].message.content)

Caveats:

• Aggregators have their own rate limits.
• Free quota is shared and can disappear fast.
• When free access ends, requests return 402 or 429. Use these for prototyping only.

Which free path should you pick?

Use case	Best free path
Terminal-based coding assistant	Codex CLI (Path 1)
Quick Python or Node experiments	Trial credit (Path 2)
Testing from a hosted app	Aggregator (Path 3)
Comparing GPT-5.5 against GPT-5.4 on real prompts	Trial credit + Apidog collection
One-off “can this answer my question” research	ChatGPT Plus (not free, but cheapest per hour)

For anything beyond a prototype, these free paths run out quickly. Use them to learn request shapes and tune prompts before switching to paid tokens.

Pre-build the request shape in Apidog

To avoid rewrites when moving from free to paid, build your request in a version-controlled Apidog collection.

In Apidog:

1. Create a new collection and add a POST https://api.openai.com/v1/responses request.
2. Set the auth header from an environment variable to swap keys without editing requests.
3. Save an example response so downstream developers can work with mocks when keys are missing.
4. Clone the collection for aggregator testing: point baseUrl at OpenRouter and change the model string.

When trial credit expires or you upgrade, just update the environment variable—no other changes required. For VS Code integration, see our Apidog in VS Code walkthrough.

Free-path limitations to plan around

• Rate limits vary by load: Codex Free/Go slows down during peak hours.
• Trial credits do not stack: Duplicate accounts (same card/phone/IP) do not get multiple trials.
• GPT-5.5 Pro is always paid: No free access to Pro.
• Thinking mode uses more quota: Keep reasoning.effort at low unless you need accuracy.
• Free windows are temporary: Codex Free/Go access is time-limited (OpenAI launch announcement).

A realistic free-tier prototype workflow

1. Pick a real team task: report, code review, research brief, etc.
2. Run 10 examples through GPT-5.4 (existing tools), log output quality.
3. Run the same 10 examples through GPT-5.5 (Codex CLI or trial credit).
4. Compare output-per-token and error rate.
5. Decide if the upgrade justifies the extra cost for your workload.

This can be done in an afternoon, and will inform your production choice before committing to paid usage.

FAQ

Is Codex Free/Go trial permanent?

No. OpenAI calls it “limited time.” Expect it to end within months.

Does ChatGPT Free provide GPT-5.5 in-browser?

No. Free ChatGPT uses GPT-5.3. GPT-5.5 requires Plus or higher.

Can I run GPT-5.5 for free via Hugging Face or Ollama?

No. GPT-5.5 is closed-weight; only available via OpenAI or Codex sign-in.

Is there a student discount?

OpenAI has offered .edu email perks before. Check the OpenAI education page for current offers.

How do I switch from free to paid without rewriting code?

Use environment variables for OPENAI_API_KEY and OPENAI_BASE_URL. When your trial ends, just change these. See our GPT-5.5 API guide for best practices.