ChatGPT Pricing in 2026: Every Plan, Tier, and Hidden Cost Explained

Quick Verdict:

ChatGPT now runs on a seven tier ladder, from a free plan up to a $200 per month Pro plan, with a $20 Plus plan sitting where most people should land. Plus is the first tier that unlocks the GPT-5.6 flagship model, the ChatGPT Work agent, Sites, and full Codex access. Go at $8 is a real budget option but it does not get you the flagship model. Pro at $100 or $200 buys usage headroom, not extra features.

OpenAI has spent the past few months quietly rebuilding both its model lineup and its price list, and the two changes are tangled together in a way that is easy to miss.

The GPT-5.6 family landed in ChatGPT on July 9, with three models carrying names instead of numbers. The macOS ChatGPT app and the Codex app merged into one desktop app. A new mode called ChatGPT Work shipped. A hosted website builder called Sites appeared. And a second Pro tier showed up at half the price of the old one.

So the obvious question is worth asking again from scratch: how much does ChatGPT actually cost, and what does each tier really buy you?

In this guide I break down every current plan, explain which models each one gets (this is where most pricing guides go wrong), and help you figure out where the money stops being worth it.

What ChatGPT Is Now, and What Just Changed

For most of its life, ChatGPT was a chat box with a model behind it. That is no longer a fair description.

The July release turned it into something closer to a work environment. The rollout started globally on July 9, 2026 and covered ChatGPT, ChatGPT Work, Codex, and the API at the same time.

The GPT-5.6 model family: Sol, Terra, and Luna

OpenAI changed how it names models. The number now identifies the generation, while Sol, Terra, and Luna identify durable capability tiers that can advance on their own cadence.

• Sol is the flagship, built for hard reasoning, long agentic runs, and complex coding
• Terra is the everyday workhorse, described as competitive with GPT-5.5 while being roughly half the cost
• Luna is the fast, cheap tier for high volume work

Here is the part that trips people up. In standard ChatGPT conversations, GPT-5.5 Instant remains the default for fast everyday responses. GPT-5.6 Sol powers the reasoning options on eligible plans, GPT-5.6 Sol Pro powers Pro, and Terra and Luna are not selectable in normal chat at all. They live in ChatGPT Work and Codex.

So “getting GPT-5.6” does not mean every message you send suddenly runs on it.

Reasoning effort, and the new ultra mode

Each model exposes reasoning effort settings, and there is a new top setting called ultra that only Sol unlocks. Ultra coordinates multiple agents across parallel workstreams, using four agents by default, trading much higher token use for stronger results on complex tasks.

That matters for your wallet, not just your output quality. Higher effort settings burn through usage allowances dramatically faster. This is the single most underrated cost lever in the current lineup.

The app merge, Work, and Sites

The Codex app and the ChatGPT desktop app are now one application. The launch bundled ChatGPT Work, a new desktop app merging Codex and ChatGPT, and a Sites beta. Work and Codex look similar, with Codex tuned for coding and Work tuned for everything else.

Sites lets you build hosted websites from inside ChatGPT, with an optional “Login with ChatGPT” feature. It is a Plus and above feature, not available on Free or Go.

ChatGPT Pricing: Every Plan at a Glance

Seven tiers, two of which confusingly share the name “Pro” at different price points.

Plan	Price	Flagship Model (GPT-5.6 Sol)	Reasoning Context	Best For
Free	$0	No	Varies	Casual questions, trying the product
Go	$8 / month	No	256K	Budget users who want volume, not frontier models
Plus	$20 / month	Yes	256K	Most individual professionals
Pro (5x)	$100 / month	Yes, plus Sol Pro	400K	Users who hit Plus caps weekly
Pro (20x)	$200 / month	Yes, plus Sol Pro	400K	Users who hit Plus caps daily
Business	$20 / user / month (annual), $25 monthly	Yes	Expanded	Teams of 2 or more
Enterprise	Custom	Yes, plus Sol Pro	Expanded	Large organizations with compliance needs

Note on the two Pro tiers: they are not the same plan billed differently. Both give you the same models. The difference is usage allowance, 5x Plus limits at $100 and 20x Plus limits at $200. Check the exact plan label in billing before you subscribe, because it is easy to click into the wrong one.

1. Free ($0)

The free tier is a genuine product, not a demo. It just has a hard ceiling and it does not get you the flagship model.

What you get

• Limited access to GPT-5.5 Instant, the fast everyday default
• Limited messages, uploads, and image generation
• Limited deep research and limited Codex access
• Limited ChatGPT Work access on the desktop app
• Search, Canvas, Study mode, Projects, and Skills beta
• 27K total context on the Instant model, roughly 12 pages of text as input

Limitations

• No GPT-5.6 Sol, and no Sol Pro
• No Sites, no scheduled tasks, no custom GPT creation
• Response times are limited by bandwidth and availability
• Your content may be used for model training unless you opt out
• OpenAI’s plan cards carry an ads notice on the lower tiers, and multiple trackers report ads running on Free and Go in the US

2. Go ($8/month)

Go went global in January and sits between Free and Plus. It is the tier I would think hardest about, because the price gap to Plus is small and the feature gap is large.

Included features

• More GPT-5.5 Instant access, more messages, more uploads
• Longer memory and memory with past chats
• 256K reasoning context, roughly 320 pages of text as input
• Voice with video, file uploads, data analysis, vision
• Scheduled tasks and the ability to create and share custom GPTs
• Limited GPT-5.6 Terra access inside Work and Codex on desktop

Limitations

• No GPT-5.6 Sol in normal chat, which is the whole point of the new generation
• No Sites, no record mode, no interactive tables and charts
• Deep research is limited rather than full
• May include ads

Go makes sense if $8 is a hard ceiling. If you can stretch to $20, the jump is not incremental, it is a different product.

3. Plus ($20/month)

This is the plan most people should buy, and the price has not moved in years despite several model generations passing through it.

Plus is the first tier where GPT-5.6 Sol appears in the model picker. Manually choosing Medium, High, or Extra High uses GPT-5.6 Sol on eligible plans.

Included features

• GPT-5.6 Sol through the medium and higher effort settings
• GPT-5.6 Terra and Luna inside ChatGPT Work and Codex
• Expanded ChatGPT Work access across desktop, web, and mobile
• Expanded Codex usage, plus the ability to buy extra credits when you hit the cap
• Sites, record mode, developer mode beta, interactive tables and charts
• Image generation with Thinking, expanded deep research, legacy models
• 256K reasoning context and 54K Instant context
• Early access to new features

Limitations

• No Sol Pro, which is reserved for the Pro tiers
• Usage is expanded, not unlimited, and high effort settings drain it fast
• No admin tools, and no training exclusion by default
• No company knowledge or workspace agents (those are Business and Enterprise features)

For a freelancer, analyst, marketer, or developer using ChatGPT daily, Plus is where the value curve is steepest.

4. Pro ($100 or $200/month)

Pro is where the pricing gets genuinely confusing, because two different plans wear the same badge.

Both Pro tiers give you the same model access, including GPT-5.6 Sol Pro, the highest capability option for difficult tasks and long running workflows. Both give you 128K Instant context and 400K reasoning context, roughly 680 pages of text as input. Both give you maximum Codex tasks, maximum deep research, unlimited and faster image creation, and research previews of new features.

What separates them is volume:

Pro at $100

• Roughly 5x Plus usage limits
• The rational upgrade if you hit Plus caps a few times a week
• Priced deliberately against Anthropic’s Claude Max 5x tier

Pro at $200

• Roughly 20x Plus usage limits
• Ultra mode access in ChatGPT Work, which Plus does not get
• Worth it only if you exhaust Plus daily, or run parallel agentic workloads for hours

If you are guessing that you “might” need Pro, you probably do not. Run a normal week on Plus and count how many times a limit actually stopped you working. If the answer is zero, stay put.

ChatGPT for Teams: Business and Enterprise Plans

Once you are buying for a group, the plans change shape entirely. The old Team plan was renamed and repriced earlier this year.

Business: $20 per user per month

Business is $20 per user per month billed annually, or $25 per user per month billed monthly, with a minimum of 2 users. That two seat floor means the practical entry point is around $40 per month.

What the seat buys you beyond Plus:

• Unlimited core chat and access to the best models for work
• 60 or more apps that bring your tools and data into ChatGPT, including Slack, Google Drive, SharePoint, GitHub, and Atlassian
• Company knowledge, workspace agents in research preview, and shared projects
• Admin console, bulk member management, admin roles, unified billing
• SAML SSO, MFA, domain verification, SOC 2 Type 2 compliance
• No training on your data by default

For agencies and content teams handling client material, the training exclusion alone is usually the deciding factor.

Business Codex: usage based, no seat fee

There is now a separate development focused plan. Business Codex carries no fixed seat fee and you pay as you go based on usage, with automated code and security reviews, cloud environments for multi agent workflows, and the same SSO and privacy guarantees.

This is a meaningful structural change. For a team with three heavy Codex users and eight occasional ones, usage based billing can land well below eleven full seats.

Enterprise: custom pricing

Enterprise adds the things regulated buyers actually ask about:

• SCIM, Enterprise Key Management, user analytics, domain verification, and role based access controls
• Custom data retention policies and data residency support in ten regions
• 24/7 priority support, SLAs, custom legal terms, and volume discounts
• Ultra mode access in ChatGPT Work
• ISO 27001, 27017, 27018, and 27701 certification

You will need to talk to sales. There is no published number, and there will not be one.

Worth knowing: OpenAI offers a free ChatGPT for Teachers plan for verified US K-12 educators through June 2027, and nonprofits can access up to a 75% discount on Business or Enterprise.

How ChatGPT’s Pricing Compares to Other AI Tools

The mid tier has converged almost perfectly across providers. The interesting differences are at the top and the bottom.

Tool	Entry Paid Tier	Power Tier	Standout	Weak Spot
ChatGPT	Go $8, Plus $20	Pro $100 or $200	Widest tool surface: Work, Codex, Sites, custom GPTs	Two plans share the “Pro” name, model access varies by tier in ways the plan cards do not make obvious
Claude	Pro $20	Max $100 or $200	Strong on long form writing and coding, agentic desktop tooling included at every tier	No sub $20 paid tier
Google Gemini	Google AI Pro $19.99	Google AI Ultra, restructured downward at I/O 2026	Very large context, deep Gmail and Docs integration, storage bundled in	Value drops sharply if you are not in the Google ecosystem
Grok	SuperGrok $30	SuperGrok Heavy $300	Live social data from X, distinctive tone	Most expensive entry tier, thinnest productivity tooling
Perplexity	Pro $20	Max $200	Sourced research with citations	Search product, not a general assistant

Where ChatGPT wins on price: the $8 Go tier has no real equivalent at Claude or Anthropic, and the $100 Pro tier closed a gap that used to force a 10x jump from $20 to $200.

Where it loses: the plan structure is now the most complicated in the category, and OpenAI’s own pricing page does not make it obvious that Go does not get the flagship model.

Competitor pricing has moved repeatedly this year, particularly Google’s, which restructured its top tier at I/O 2026. Verify current numbers on each provider’s own page before committing.

What Do You Actually Get with ChatGPT?

The price list tells one story. Here is what the day to day experience looks like on a paid tier.

Core capabilities

• GPT-5.5 Instant handles fast everyday questions by default, with Sol available when you select a reasoning level
• Deep research runs multi step investigations across dozens of sources
• Memory carries context across conversations, and memory with past chats is expanded on paid tiers
• Search, Canvas, data analysis, and vision are available on every tier including Free

The agentic layer

• ChatGPT Work turns source material from documents and connected apps into shareable outputs
• Codex handles coding tasks, from local edits to cloud tasks that open pull requests
• GPT-5.6 improves artifact quality across presentations, documents, and spreadsheets, with outputs exportable into existing enterprise tools
• Computer use lets the model drive your cursor, open apps, and click through interfaces by looking at the screen

What you do not get

• API access is a completely separate bill. A Plus, Pro, or Business subscription includes zero API credits
• Workspace agents and company knowledge are Business and Enterprise only
• Ultra mode in ChatGPT Work is Pro and Enterprise only

The Numbers That Actually Constrain You

Most pricing guides list features. The limits are what you will actually run into. These are the published figures, not estimates.

Metric	Figure
GPT-5.5 Instant messages, Plus and Go	Up to 160 every 3 hours, then chats switch to Instant mini
Thinking messages on Go	Up to 10 every 5 hours, and this option does not use Sol
Instant context window	27K Free, 54K Go and Plus, 128K Pro
Reasoning context window	256K Go and Plus, 400K Pro
Pro usage multipliers	5x Plus at $100, 20x Plus at $200
Ultra mode agent count	Four agents in parallel by default
API pricing per 1M tokens	Sol $5 in / $30 out, Terra $2.50 in / $15 out, Luna $1 in / $6 out
Codex overage	Plus and Pro users can buy additional credits without upgrading

The most important line in that table is not a price. It is the fact that higher reasoning effort settings consume your allowance far faster. Running everything at max or ultra is the quickest way to exhaust a plan you thought was generous.

Is ChatGPT Worth It?

It depends entirely on which tier you are asking about.

Plus at $20 makes sense if you

• Use ChatGPT for real work most days rather than occasional questions
• Want the flagship model, not the everyday default
• Need deep research, Codex, Sites, or the Work agent
• Care about response reliability during peak hours

Go at $8 makes sense if you

• Use ChatGPT regularly but only for straightforward tasks
• Keep hitting free tier message caps
• Do not need frontier reasoning, and can live with ads

Pro makes sense if you

• Can name specific moments in the last month when Plus limits stopped you working
• Run Codex or agentic workflows for hours at a time
• Need Sol Pro or ultra mode specifically, not just “the best plan”

Skip the upgrade if you

• Are buying on the assumption you will grow into it
• Mainly need one capability that a $20 competitor does better
• Are a solo user considering Business, which adds nothing for one person that Plus or Pro does not already cover

From my perspective, Plus at $20 is still where the real value sits, and the $100 Pro tier is the genuinely useful addition to the lineup because it removes the old all or nothing jump to $200.

Recent Changes Worth Knowing About

If you are reading a ChatGPT pricing guide written even three months ago, parts of it are already wrong. Here is what actually moved.

Change	When	Why It Matters
Go plan went global at $8	January	Created a real budget tier below Plus
Business repriced and renamed from Team	April 2	Dropped to $20 per seat annual, $25 monthly
Second Pro tier launched at $100	April 9	Closed the gap between $20 and $200
GPT-5.6 Sol, Terra, and Luna reached ChatGPT	July 9	New flagship on Plus and above, new naming system
Codex app merged into the ChatGPT desktop app	July 9	One app, with Work and Codex as modes
ChatGPT Work and Sites shipped	July 9	Agentic work mode and hosted website building, Plus and above
Ultra mode introduced	July 9	Parallel subagents on Sol, Pro and Enterprise in Work

Two practical warnings from the transition. First, the app merge shipped with bugs, and OpenAI reset usage allowances several times while fixing them. Second, the temporary removal of some rolling window limits means it is genuinely possible to burn a weekly allowance in a single session if you leave the effort setting high.

Final Thoughts on ChatGPT Pricing

ChatGPT’s price list has gone from “free or twenty dollars” to a seven rung ladder in about eighteen months, and the complexity is not accidental. OpenAI is segmenting hard, and the segmentation is now driven by usage volume and model access rather than by features.

If you are new, start on Free. It is a real product.

If ChatGPT is part of how you work, Plus at $20 is the answer and has been for three years. It is the first tier that gets you GPT-5.6 Sol, the Work agent, Codex, and Sites, and the feature gap between it and Go is far larger than the $12 price difference suggests.

Pro is a usage purchase, not a capability purchase. Buy it when a limit has actually stopped you working, not before. And if you are running a team, Business at $20 per seat buys the one thing individual plans cannot: your conversations stay out of the training data by default.

The models will keep changing. The plan structure probably will too. Check the official pricing page before you commit, because in this category a guide is stale the month it is published.