OpenAI API Pricing in 2026 — What GPT Costs per Request, Model by Model

The OpenAI API is billed per token, not per month. Input and output have separate rates, and output costs more. On Kunavo, the range runs from $0.04 / $0.20 per million tokens for GPT-6 Luna to $4.00 / $20.00 for GPT-6 Astra, OpenAI’s top-end model; between them are GPT-6 Sol at $0.80 / $4.00 and the workhorse GPT-5.6 Terra at $0.70 / $4.20 — 60 to 65% below OpenAI's list price, depending on the model. A ChatGPT subscription is a separate bill: it does not include the API, and the API does not require a subscription.

Prices checked on October 3, 2026. This page reads Kunavo prices live from the model catalog: what is displayed is what the API charges. The “OpenAI pricing” column uses the price published by OpenAI for the same model at openai.com/api/pricing. Kunavo is a gateway that resells access to the models, not an independent comparison service.

OpenAI API prices by model

In dollars per million tokens, as charged by Kunavo. The “OpenAI pricing” column is the public price for the same model.

Live prices are always on the pricing page and on each model page, for example GPT-6 Sol or GPT-6 Luna. Which model for what:

• gpt-6-sol — positioned by OpenAI for complex code and agent workflows. On Kunavo, $0.80 / $4.00, below GPT-5.6 Terra on output, slightly higher on input. The reasonable default choice for code and agents.
• gpt-6-luna — according to OpenAI, the most efficient model for focused, high-volume tasks: classification and extraction. Kunavo’s cheapest GPT.
• gpt-6-astra — the top end: cutting-edge reasoning and long agent sessions. For the difficult minority of tasks.
• gpt-5-6-terra — the workhorse of the 5.6 family: RAG, assistants, and most production tasks.
• gpt-5-6-sol and gpt-5-5 — previous generations. On Kunavo, both cost the same per token; moving from gpt-5-5 to the next generation depends solely on the model name.

“Prices cut in half”: compared with what?

The press presented the launch of GPT-6 Sol and Luna as a price reduction. At OpenAI, GPT-6 Sol currently costs $2.00 / $10.00 per million tokens versus $4.00 / $20.00 for GPT-5.6 Sol (OpenAI promotional rate, at least until November 21, 2026; standard rates $5.00 / $30.00): 2 times less for input and 2 times less for output. It is not the price of the same model that fell; it is a new, cheaper model — and GPT-6 Astra, the top-end model, remains at $10.00 / $50.00. If your code still calls gpt-5-6-sol, changing the model name is the simplest price reduction to obtain; check quality on your own inputs before switching.

GPT-6 Sol and GPT-6 Luna have the same context window, 1,050,000 tokens, and return up to 128,000 output tokens. A surcharge for very long requests: above 272,000 prompt tokens, GPT-5.5, the GPT-5.6 family, and the three GPT-6 models — Astra, Sol, and Luna — bill the entire request at 2× the input rate and 1.5× the output rate.

ChatGPT subscription and OpenAI API: two products, two bills

Anyone searching for “ChatGPT API pricing” often first wants to know whether their subscription already covers the API. It does not: a subscription provides neither an API key nor API credit, and API credit does not increase ChatGPT limits. Subscription prices are at openai.com/chatgpt/pricing; they change often, so they are not reproduced here. For someone who spends the day in the chat window, the subscription is often the least expensive option: an exchange of 2,000 input tokens and 700 output tokens costs approximately $0.0044 on GPT-6 Sol and $0.00434 on GPT-5.6 Terra; divide the plan price by that amount to get the threshold. The API wins for software, automation, and irregular usage, where a quiet month should cost nothing.

What a request really costs

At Kunavo prices, without caching and reasoning tokens — therefore the floor of the bill:

The minimum $10 top-up pays for approximately 1,590 RAG responses on GPT-5.6 Terra. Two costs every estimate misses: reasoning tokens, billed at the output rate and absent from the response text, and retries — a response received and then discarded by your code is still paid for. Requests that fail at Kunavo are not billed. For savings, the prompt cache bills reread tokens at a fraction of the input rate: $0.20 instead of $2.00 per million on GPT-5.6 Sol (see the cache documentation).

Paying for the API from France

Kunavo bills through Stripe, using a prepaid dollar balance: no subscription, no OpenAI platform account to configure. At checkout:

• Visa, Mastercard, American Express, JCB, and UnionPay cards — a debit card linked to your SEPA account normally works
• Apple Pay, Google Pay, and Link

SEPA direct debit is not offered. Pay by Bank is not available either: Kunavo's checkout displays it only to buyers in the United Kingdom, Ireland, and Finland, and Stripe offers it in France only in private preview (Stripe documentation on Pay by Bank, read on October 3, 2026). Stripe displays the amount in euros at checkout. The minimum top-up is $10; above that, a bonus is added: $100 becomes $110.00, $1,000 becomes $1,200.00, and $5,000 becomes $6,250.00. The balance never expires. Details are in the billing documentation.

Move your OpenAI code to Kunavo

Kunavo is OpenAI-compatible: only the base_url and the key change. The response contains the usual usage field, which lets you recalculate every invoice line.

from openai import OpenAI

client = OpenAI(
    api_key="sk-kn-...",
    base_url="https://api.kunavo.com/v1",  # au lieu de api.openai.com
)

resp = client.chat.completions.create(
    model="gpt-6-sol",
    messages=[{"role": "user", "content": "Résume cette facture en trois phrases."}],
)
print(resp.choices[0].message.content)
print(resp.usage)  # tokens d'entrée, de sortie et de raisonnement : la base de la facture

The key (sk-kn-…) is created in the dashboard after sign-up; you can set a monthly spending limit per key there. The steps are in the quickstart.

When to go directly through OpenAI

Kunavo is a gateway with shared capacity: there is no quota dedicated to your account and no contractual SLA. For a guaranteed quota or an SLA — a product whose availability is promised to your customers — buy directly from OpenAI. To recalculate the figures with your numbers, the OpenAI API pricing calculator accounts for cached input and reasoning tokens. The Anthropic equivalent is Claude API pricing.

Frequently asked questions