What Does the ChatGPT API Cost Per Month in 2026?

Short answer: The ChatGPT API can cost $0.15 a month for 100 short messages on GPT-5.6 Luna, about $13.20 for 300 everyday messages on GPT-5.6 Terra, or $96 for 600 long messages on GPT-5.6 Sol. Those figures are token usage only. Add a $4 Multimodel Chat workspace fee if you want a BYOK chat interface. ChatGPT Plus is a separate $20/month product and does not include API credits.

The phrase "ChatGPT API cost per month" sounds like it should have one answer. It does not. OpenAI bills the API by input and output tokens, so the monthly total moves with your model, message length, context window, and tool use. A person sending short prompts a few times a week can spend less than a dollar. A developer sending long documents through a premium model can pass $20 in one afternoon.

This guide uses current OpenAI standard rates and three reproducible usage profiles. The numbers are meant to give you a starting point, not a promise about your bill. Put your own message count and token estimates into the live cost calculator before you commit to a setup.

ChatGPT API cost per month: the simple formula

OpenAI lists prices per 1 million tokens. The basic calculation is:

monthly cost = (input tokens / 1,000,000 × input rate)
              + (output tokens / 1,000,000 × output rate)

Multiply that result by your number of messages. Input tokens include the prompt and any conversation history you send. Output tokens are the model's response. A long conversation can make the input side much larger than the words you typed in the latest message.

API prices are separate from ChatGPT Plus. OpenAI's current help page lists Plus at $20/month, billed monthly, and says API usage is billed independently. Paying for Plus does not create an API key or give you API credits. If you want a chat interface around your own OpenAI key, you pay the provider for tokens and the interface provider, if any, for the workspace.

OpenAI API prices to use in 2026

The table below uses the standard short-context rates on OpenAI's live pricing page. Prices are per 1 million tokens.

Model	Input	Cached input	Cache write	Output
gpt-6-astra	$10.00	$1.00	$12.50	$50.00
gpt-5.6-sol	$4.00	$0.40	$5.00	$20.00
gpt-5.6-terra	$2.00	$0.20	$2.50	$12.00
gpt-5.6-luna	$0.20	$0.02	$0.25	$1.20

The pricing page says GPT-5.6 Sol's promotional $4 input and $20 output rates are available at least through November 21, 2026. Check the page again after that date. Cached input is cheaper than normal input, but you only get that rate when OpenAI recognizes a cache hit. Cache writes have their own price. The profiles below use standard input rates and do not assume a cache discount, so they are easier to reproduce.

These are model-token charges. Image inputs, audio, web search, batch processing, regional processing, and other features can add separate costs or use different rates. The API also has context limits. A request that fails before inference may not cost the same as one that completes, but a successful request with a large history still sends those tokens for billing.

Three ChatGPT API cost profiles

Each profile uses a different model and message pattern. The arithmetic is visible so you can replace the assumptions with your own numbers.

Profile 1: light use

Assume 100 messages per month, 3,000 input tokens and 750 output tokens per message, using GPT-5.6 Luna:

• Input: 100 × 3,000 / 1,000,000 × $0.20 = $0.06
• Output: 100 × 750 / 1,000,000 × $1.20 = $0.09
• API total: $0.15/month

This is a useful profile for occasional personal questions, short writing edits, or a small side project. If you use a BYOK workspace, a $4/month Pro plan would make the combined total $4.15 before taxes and any search usage. The workspace fee becomes the larger line item, which is normal for light API use.

Profile 2: everyday work

Assume 300 messages per month, 10,000 input tokens and 2,000 output tokens per message, using GPT-5.6 Terra:

• Input: 300 × 10,000 / 1,000,000 × $2.00 = $6.00
• Output: 300 × 2,000 / 1,000,000 × $12.00 = $7.20
• API total: $13.20/month

Add the $4 Multimodel Chat Pro workspace fee and the total is $17.20. That is below ChatGPT Plus's $20 subscription in this specific scenario, while also leaving the provider invoice separate and letting you connect other providers. It is not an apples-to-apples feature comparison: Plus includes ChatGPT product features and usage limits that a plain API key does not.

Profile 3: heavy long-context use

Assume 600 messages per month, 20,000 input tokens and 4,000 output tokens per message, using GPT-5.6 Sol:

• Input: 600 × 20,000 / 1,000,000 × $4.00 = $48.00
• Output: 600 × 4,000 / 1,000,000 × $20.00 = $48.00
• API total: $96.00/month

With a $4 workspace fee, the total reaches $100. In this pattern, ChatGPT Plus is cheaper on the headline price, assuming its limits and features suit your work. This is why any answer that calls the API "cheap" without stating tokens is incomplete.

Profile	Model	Messages	Tokens per message (in/out)	API usage	With $4 workspace	ChatGPT Plus
Light	GPT-5.6 Luna	100	3,000 / 750	$0.15	$4.15	$20
Everyday	GPT-5.6 Terra	300	10,000 / 2,000	$13.20	$17.20	$20
Heavy	GPT-5.6 Sol	600	20,000 / 4,000	$96.00	$100.00	$20

API or ChatGPT Plus?

The prices answer only one part of the decision.

If you care about...	ChatGPT Plus	OpenAI API with BYOK
Billing	Fixed $20/month	Metered usage
API access	Not included	The product itself
Other model providers	Not in the same workflow	Possible through a compatible workspace
Product features	ChatGPT's bundled tools	Depends on the client you choose
Spend control	Predictable bill, plan limits	Set provider budgets and monitor usage
Setup	Sign up and use it	Create a key, set billing, choose a client

Choose Plus if you want one consumer product, a predictable bill, and ChatGPT-specific tools. Choose the API if you are building software, need direct programmatic access, use models selectively, or want a workspace that combines providers. Read the existing ChatGPT Plus vs API comparison for the product-level tradeoffs, then use the OpenAI setup guide when you are ready to create a key.

Costs that the simple math misses

A token estimate can still undershoot your invoice.

• Conversation history: Sending the full thread on every turn increases input tokens. A 20,000-token prompt is not the same as a 500-token question.
• Search and tools: A model call can trigger web search, code execution, or another paid service. Count those calls separately.
• Retries: Timeouts and application retries can repeat a request. Add request IDs and a usage log before a production launch.
• Images and audio: Multimodal inputs have their own tokenization and pricing. Do not apply the text table to an image-heavy workflow.
• Workspace fees: A chat app can charge for the interface while OpenAI charges for inference. Keep the two lines separate.

Set a monthly budget on your OpenAI project before you start. Use a low-spend test key for experiments, and watch the usage dashboard during the first week. A hard cap is more useful than trying to remember which conversation became expensive.

Which should you choose?

If you...	Best fit
Send fewer than 100 short messages a month	API, if you are comfortable with setup
Want a fixed personal bill and ChatGPT features	ChatGPT Plus
Need API access for an application	OpenAI API
Use GPT, Claude, Gemini, or Grok in one workflow	BYOK workspace plus provider keys
Send long contexts to a premium model every day	Calculate carefully; Plus may cost less

There is no universal winner. The useful comparison is your actual token pattern plus the features you would otherwise buy separately.

FAQ

Is the ChatGPT API included with ChatGPT Plus?

No. ChatGPT Plus and the OpenAI API are separate products and billing systems. Plus is a $20/month subscription for the ChatGPT app; API calls use a funded API account and are billed by token.

What is the cheapest OpenAI model for monthly API use?

Among the models used in this guide, GPT-5.6 Luna has the lowest standard short-context rates at $0.20 per million input tokens and $1.20 per million output tokens. Your total still depends on how many tokens you send.

Can the ChatGPT API cost more than ChatGPT Plus?

Yes. The heavy profile above costs $96 in API usage before a workspace fee. Long prompts, large output limits, images, tools, and retries can push a metered bill above a flat subscription quickly.

Does a BYOK workspace change OpenAI's token rates?

It should not if the workspace passes your requests to OpenAI at provider rates. You still pay the workspace fee, and you should check its pricing and key-handling policy. Multimodel Chat lists Pro at $4/month and Lifetime Access at $39 one time, with provider usage billed separately.

---

Start your free trial → — 7 days, all providers, no credit card required.