OpenAI API pricing per 1M tokens for every current model

GPT-6 and GPT-5.6 prices per 1M tokens

OpenAI's guide to its latest GPT-6 models describes four options:

• GPT-6 Astra is OpenAI's most intelligent model, aimed at computer use, browsing, software engineering and science.
• GPT-6.1 Sol gets close to Astra for less money on coding, computer use and professional work.
• GPT-6.1 Terra is the fast, efficient option for visual reasoning and writing.
• GPT-6 Luna is the fastest and cheapest of the lineup.

The table below lists Standard tier prices per 1M tokens for prompts up to 272K input tokens. The $4 input rate on gpt-5.6-sol is promotional, available at least through November 21, 2026.

Batch, Flex, Fast and Ultrafast

The same model costs a different amount on each processing tier.

• Batch runs asynchronous jobs that finish within 24 hours, costs 50% less than Standard, and pulls from a separate pool with much higher rate limits.
• Flex bills Responses or Chat Completions requests at Batch rates when you pass service_tier set to flex, though it's in beta for a limited set of models and can return a 429 Resource Unavailable error (which isn't charged).
• Fast mode (service_tier set to fast) gives up to 2.5x faster output and was called Priority processing until July 30, 2026.
• Ultrafast costs 6x Standard and is priced only for GPT-6 Astra and GPT-6.1 Sol.

The table below compares input and output prices per 1M tokens by tier.

Long prompts and prompt caching

• Requests with more than 272K input tokens are billed at long-context rates: gpt-6-astra moves to $20 input and $75 output per 1M tokens, and gpt-6.1-sol to $4 and $15.
• Reading cached input costs a tenth of the standard input price on gpt-6-astra and gpt-6-luna, and a twentieth on gpt-6.1-sol and gpt-6.1-terra ($0.10 vs $2.00 for Sol).
• Writing to the cache on GPT-6 and GPT-5.6 models costs 1.25x the input price for a 30-minute cache and 2.5x for a 12-hour cache.
• Data residency (regional processing) and FedRAMP endpoints add 10% for models released on or after March 5, 2026.

Older and smaller models

OpenAI still lists several earlier models on its pricing page. Pro variants cost far more than the base models: gpt-5.5-pro is $30 input and $180 output per 1M tokens, and o1-pro is $150 and $600.

The table below shows the earlier models at Standard tier rates per 1M tokens.

Images, audio, embeddings and tools

• gpt-image-2 costs $8 per 1M image input tokens, $5 per 1M text input tokens and $30 per 1M image output tokens (see our per-image cost breakdown).
• Embeddings cost $0.02 per 1M tokens on text-embedding-3-small and $0.13 on text-embedding-3-large.
• Realtime audio on gpt-realtime-2.1 runs $32 input and $64 output per 1M tokens.
• Audio transcription starts at about $0.003 a minute with gpt-4o-mini-transcribe and $0.0045 a minute with gpt-transcribe.
• Built-in web search costs $10 per 1,000 calls plus search content tokens, and file search storage is $0.10 per GB per day after the first free GB.

How OpenAI API billing works

When you set up a new account and grab an OpenAI API key, you're on prepaid billing. The minimum credit purchase is $5 (the default is $10), and any credits you buy expire after 1 year and are non-refundable unless law, contract or an approved exception requires a refund.

ChatGPT subscriptions are billed separately from the API, so there's no free OpenAI API usage included with ChatGPT Plus, Go or Free. Paid Pro 100, Pro 200 and Pro 500 plans can claim $100, $200 or $500 of monthly API credits, and Business plans get $20 per paid Standard seat or $100 per paid Premium seat (pooled and capped at $500 a month per workspace), though those credits don't cover Codex used through API keys.

Usage tiers cap monthly API usage and rise automatically with your total credit purchases:

• Free tier gives users in an allowed geography a $100 monthly usage limit.
• Build tier raises the limit to $500 a month after $5 in total credit purchases.
• Launch tier raises the limit to $5,000 a month after $100 in purchases.
• Grow tier goes up to $200,000 a month once you've bought $500 in credits.

Pay with a ChatGPT plan instead of per token

If you already pay for ChatGPT Plus or Pro, Plan2API lets you use your ChatGPT subscription as an API by connecting it through Codex and giving you a single p2a_ key for https://api.plan2api.com/v1. You call standard OpenAI Chat Completions with Codex models named openai/<model> like openai/gpt-6.1-sol, your requests run against the plan's own limits, and Plan2API never bills per token.

The trade-off is endpoint coverage: Plan2API serves POST /v1/chat/completions, GET /v1/models and, on Pro and Scale plans, the image endpoints (POST /v1/images/generations and POST /v1/images/edits), so you don't get the /v1/responses API, the Batch API or /v1/embeddings. Plan2API's Free plan includes 50 chat requests a day for $0, and Pro is $12 a month with no daily request cap.

import OpenAI from "openai";

const client = new OpenAI({
  baseURL: "https://api.plan2api.com/v1",
  apiKey: process.env.PLAN2API_KEY,
});

const completion = await client.chat.completions.create({
  model: "openai/gpt-6-luna",
  messages: [{ role: "user", content: "Classify this ticket: 'refund not received'" }],
});

console.log(completion.choices[0].message.content);