ChatGPT Pricing — plans, limits and API costs

Plans at a glance

Go, Plus and Pro are monthly only; there's no annual billing (OpenAI). Education options: ChatGPT for Teachers (free for verified US K–12 educators through June 2027) and ChatGPT Edu. Nonprofits can get up to 75% off Business or Enterprise.

Limits worth knowing before you pick

• "Unlimited" means everyday text chats. Tools like uploads, image generation, voice, data analysis, reasoning, Work and Codex have their own allowances on every plan (Free tier FAQ, Business limits).
• ChatGPT Work and Codex share one allowance, then optional credits. GPT-6 Astra uses it faster. Plus and Business Standard get limited Astra usage; Pro and Business Premium can spend their whole allowance on it.
• Business Pro messages: Standard seats get 15 Pro messages a month and Premium seats 50 a week, shared across GPT-6 Pro and GPT-5.6 Sol Pro.
• Active scheduled tasks: 3 on Free/Go, 5 on Plus, 10 on Business/Edu, 15 on Pro/Enterprise.
• Custom instructions: 1,500 characters on Free and Go, 5,000 on paid plans.
• Custom GPTs: personal plans can use existing GPTs but not create new ones, and GPTs retire on Dec 11, 2026. See Custom GPTs → Plugins.

Which plan?

• Free or Go: everyday chat and light tool use. Go is the cheapest way to raise upload and image limits.
• Plus: the first plan with GPT-6 reasoning, Projects, scheduled tasks, event-triggered Work tasks and ChatGPT Work across all devices.
• Pro $100: if you keep hitting Plus limits or want the full Work/Codex allowance on GPT-6 Astra.
• Business: for two or more people. You get admin controls and no training on your data by default, and you can mix Standard and Premium seats. Sharing one personal login breaks OpenAI's terms. See Teams & Business.

The API is billed separately

A ChatGPT subscription doesn't include API usage. The API is pay-as-you-go per token, and some built-in tools add a per-call or per-session charge. Rates below were checked against OpenAI's pricing page on 2026-09-24:

• GPT-6 Astra (flagship, 2026-09-03): $10 / 1M input, $1 cached input, $50 output. 1,050,000-token context, 128k max output.
• GPT-6 Sol: $2 input, $0.20 cached, $10 output. GPT-6 Luna: $0.10 input, $0.01 cached, $0.50 output. Both launched 2026-09-22; long-context requests bill at a higher tier (Sol $4/$15).
• GPT-5.6 (previous): Sol $4/$20 (marked promotional through 2026-11-21) · Terra $2/$12 · Luna $0.20/$1.20.
• GPT-Live 1 (full-duplex voice): $0.05 per minute.
• Web search tool: $10 per 1,000 calls on reasoning models plus the result tokens at the model's input rate; $25 per 1,000 calls on non-reasoning models, with result tokens free.
• Code interpreter: $0.03 per 20-minute session for a 1 GB container (4 GB $0.12, 16 GB $0.48, 64 GB $1.92). File search: $2.50 per 1,000 calls plus $0.10/GB per day of storage after the first free GB.

Always confirm on OpenAI's API pricing page before budgeting. Model your own mix in our cost calculator.

Worked API examples

• Search-backed support bot, 5 queries a day on GPT-6 Sol with one web search each: about $0.02–0.025 per query (the $0.01 search fee plus a few thousand result tokens), so roughly $3–4 a month. The result tokens, not the call fee, grow with heavier searching.
• PR review bot, 10 PRs a day, one 1 GB code-interpreter session each: $9 a month in sessions plus about $1.50 in tokens, so about $10.50 a month.

How to keep ChatGPT and API costs down

• Use a subscription for people and the API for software. Don't rebuild the ChatGPT app on the API just to save a subscription.
• Point routine API work at GPT-6 Luna or Sol, and keep Astra for calls that need it.
• On reasoning models, watch search-result tokens rather than the search fee.
• Set a monthly budget in your OpenAI API organisation's billing settings so a runaway integration can't surprise you.