Anthropic Claude and OpenAI ChatGPT Individual Plans Compared

US prices in USD • Updated 10 October 2026 • For advanced users and developers

Claude and ChatGPT now sell more than access to a chatbot. Their individual subscriptions also include coding agents, research tools, file creation, and increasingly autonomous work. Choosing a plan therefore means deciding which workflows you need, which models power them, and how much interruption you can tolerate.

For most developers, Claude Pro and ChatGPT Plus are the sensible starting points at $20 per month. Claude Max deserves particular attention if you also use the Claude API, because it now bundles monthly API credits. ChatGPT Pro offers higher agent allowances and premium reasoning; its $500 tier adds exclusive Ultrafast access among individual subscriptions. These are purchasing judgments based on the features below, rather than a claim that either provider produces better results on every task.

This comparison covers ordinary individual subscriptions only. Prices are standard US web subscription prices before applicable tax; regional and app-store prices can differ. Annual figures below are calculated costs at today's prices, not guarantees against future changes. Model access and recently announced features can arrive gradually.

Individual plan prices at a glance

Claude Pro's $200 annual charge works out to $16.67 per month, although the pricing page rounds its advertised monthly equivalent to $17. It saves $40, or approximately 16.7%, against twelve $20 payments. Max is monthly only. OpenAI currently offers no annual billing for Go, Plus, or Pro. Sources: Claude pricing, Claude Max billing, US Go price, Plus price, Pro prices, and OpenAI billing terms.

Comparing Anthropic plans

Claude Free

Claude Free provides web search, memory, connectors, Artifacts, and limited use of Sonnet and Haiku. Opus and Claude Code require a paid plan. Claude individual feature matrix. Free includes up to five projects and voice; Research requires Pro or Max. Projects, voice availability, Research eligibility.

Free is already useful for technical experiments: Claude can execute code and create downloadable spreadsheets, documents, presentations, and PDFs. That computing environment is different from Claude Code working in your development environment. File generation and analysis consume the plan's usage budget, so a free subscription can demonstrate a workflow without sustaining it all day. Code execution and file creation.

Start here to evaluate Claude's writing, reasoning, and handling of your documents. Upgrade when you need its coding agent, paid models, research, or more regular access.

Claude Pro

Pro costs $20 monthly or $200 annually. It adds Claude Code, more model access including Opus, and more usage than Free. Its allowance is variable: attachments, conversation length, model choice, and task complexity all affect consumption. Pro has a five-hour session limit and a separate weekly allowance, so exhausting either can interrupt work. General API use through Claude Console is billed separately. Claude Pro terms.

The newer Claude experience merges chat and Cowork into one conversation and is rolling out first to Pro and Max. It can complete research, reports, spreadsheets, and presentations, with cloud tasks continuing after you close your laptop. Tasks involving local files or applications still require Claude Desktop to remain open. The change includes scheduling and connected-app workflows, but accounts on the same plan can receive it at different times. Unified Claude experience.

Pro is a strong first paid option for a developer who wants Claude Code alongside document work. Its weakness is capacity during sustained agent sessions. The annual discount is attractive once Claude has become a regular tool; monthly billing is easier to justify while evaluating it.

Claude Max 5x and Max 20x

Max 5x costs $100 monthly and includes five times Pro's per-session allowance. Max 20x costs $200 and includes twenty times Pro's per-session allowance. Both retain five-hour limits and weekly limits; the multipliers should not be interpreted as unlimited usage or guaranteed numbers of completed tasks. Max also offers higher output limits and earlier access to advanced features. Max plan terms, Claude pricing.

The price step from Max 5x to Max 20x doubles the subscription price while quadrupling the advertised per-session allowance. That makes the higher tier attractive for heavy usage, but only if your own workload needs the additional capacity. It does not establish a fourfold improvement in coding throughput, weekly capacity for every model, or answer quality.

As of October 2026, Max also includes monthly Claude Platform credits: $100 on Max 5x and $200 on Max 20x. New subscribers become eligible after seven days and must link a Claude Console organization. Credits refresh each billing cycle, expire without rollover, and are spent before purchased API credits. They cover direct Claude API use, Message Batches, the Playground, Managed Agents, and the Agent SDK. They do not cover interactive Claude Code, overage in the Claude apps, or Claude accessed through third-party cloud providers. The benefit is rolling out, so confirm that the claim option is available. Monthly API credit terms.

For a developer who already spends that much on eligible Claude API workloads each month, the credits can offset an existing expense. For someone who uses only interactive Claude Code, their practical value may be zero. They are restricted consumption credits, not cash, and do not make the subscription free.

Claude model access and shared limits

The current model families include Fable 5.1, Opus 5.5, Sonnet 5.5, and Haiku 5.5. The important subscription distinction is Fable: Pro pays through usage credits from the first request. Max includes Fable, but only up to 50% of its weekly allowance can be spent on Fable models, and that consumption also reduces the general allowance. Fable uses the budget faster than other models. After its ceiling, switch models or pay for usage credits. Fable plan rules, current Claude models.

Claude's usage budget is shared across its app surfaces and Claude Code. Code execution, lengthy conversations, tools, and higher effort can consume it quickly. An agent can make many model calls behind one visible request, so counting messages is a poor way to forecast capacity. Usage and length limits.

Paid users can enable usage credits to keep working after reaching included limits, including a Max Fable limit. This switches work to consumption billing at standard API rates. You can set a monthly spending cap and control automatic reload. These app overage credits are separate from Max's bundled API credits. Paid Claude usage credits.

Anthropic also offers discounted overage bundles: $50 of usage for $45, $250 for $200, and $1,000 for $700. Pro and Max users can buy up to $2,000 in face value of discounted bundles per month; further usage is charged at standard rates. These supplement included usage rather than increasing the subscription allowance. For intermittent peaks, compare a bundle with the cost of maintaining a higher tier every month. Usage bundle pricing.

Comparing OpenAI plans

ChatGPT Free

Free includes ordinary text chat, search, and limited uploads, images, voice, research, memory, and desktop agent access. OpenAI describes everyday text chats as unlimited subject to abuse safeguards, but tool use and premium-model access remain restricted. ChatGPT pricing.

On 7 October, OpenAI announced GPT-6 in ChatGPT, powered by GPT-6 Sol for paid higher tiers and GPT-6 Luna for Free and Go, with the latter rollout starting on 8 October. The same update adds visual and interactive responses through Intelligent UI. Some support text still refers to GPT-5.6 Luna, so actual account access may reflect the transition. GPT-6 rollout announcement.

Free is useful for evaluating ChatGPT and occasional development assistance. Limited Codex access should be treated as an opportunity to test the agent against your repository, rather than a full working-day allowance.

ChatGPT Go

Go costs $8 monthly in the US. It primarily expands access to tools such as uploads, image generation, and data analysis, and offers longer memory. It also lists projects, tasks, custom GPTs, and Library subject to availability. Some advanced features still require Plus or higher, so Go should not be dismissed as having no tools or presented as equivalent to Plus. Go details.

Go is a reasonable budget choice for someone who frequently analyzes files or creates images but rarely needs demanding agent work. For a developer whose main requirement is sustained Codex access, the additional $12 for Plus is likely easier to justify than regularly working around Go's constraints.

Paying for Go does not guarantee an ad-free experience. OpenAI's ads testing applies to eligible Free and Go accounts; Plus and Pro are ad-free. Turning off ad personalization does not remove ads. The Free plan's reduced-access Ads-Free option is a separate tradeoff. Ads in ChatGPT.

ChatGPT Plus

Plus costs $20 monthly and provides higher model limits, priority access, advanced reasoning, and expanded uploads, image generation, voice, and research. Limits can vary with demand. It does not include general OpenAI API usage. Plus subscription details.

For developers, the larger benefit is Codex across supported desktop, web, CLI, IDE, and mobile surfaces, plus ChatGPT Work for longer tasks. Plus includes GPT-6.1 Sol and access to Astra in Work and Codex. Astra access there does not mean Plus includes GPT-6 Pro effort in ordinary Chat. Product surface matters as much as the model name. Work and Codex access.

OpenAI currently estimates Plus local-message capacity over five hours at 5–45 for Astra, 15–160 for GPT-6.1 Sol, 15–150 for GPT-6 Sol, and 350–3,000 for GPT-6 Luna. These are workload-dependent estimates, not guaranteed message quotas. Local and cloud work share allowance, and weekly limits can also apply. Codex usage estimates.

Plus makes sense for regular coding sessions interspersed with research, file analysis, and general chat. Before upgrading further, check whether changing model, effort, or task scope makes the included allowance sufficient.

ChatGPT Pro 100 and Pro 200

The current Pro lineup starts at $100 and includes a $200 option. Both offer Pro models and more agent capacity than Plus. In ordinary Chat, Pro adds Extra High and Pro reasoning; Plus's table lists Medium and High. GPT-6 Pro remains capped, and Pro 100 shares a weekly premium-model pool between GPT-6 Pro and GPT-5.6 Sol Pro. Switching between them does not replenish it. Thinking levels and premium-model limits.

Pro 100 is reported as five times the Plus allowance, and the revised Pro 200 allowance as ten times Plus. Those numerical comparisons come from reporting of OpenAI product lead Thibault Sottiaux's announcement; the current Pro help page simply confirms that Pro 200 includes more usage than Pro 100. Treat the multiples as reported planning guidance and check your own dashboard for current limits. WinBuzzer reporting.

Pro 200 is open to new subscriptions again. New subscriptions generally receive the revised lower allowance. Eligible customers who had active Pro 200 during 22–29 September, up to 10 a.m. Pacific on the 29th, retain the previous allowance through 29 October while subscribed; afterward it drops while the price remains $200. Current Pro terms.

Unlike Plus, Pro currently has no five-hour agent limit, though longer-term and model limits still matter. For bursty workloads, that scheduling flexibility can be as important as the larger allowance. Codex usage windows.

ChatGPT Pro 500

Pro 500 costs $500 monthly. OpenAI's DevDay announcement describes it as twenty-five times the Plus allowance. It is also the only individual tier with Ultrafast access for GPT-6 Astra and GPT-6.1 Sol. Buying extra credits on lower tiers does not unlock that mode. Pro 500 announcement, Ultrafast eligibility.

Speed has a material cost: Ultrafast consumes included allowance at eight times the corresponding Standard rate and purchased credits at six times Standard. Fast consumes included allowance at 2.5 times Standard and purchased credits at twice Standard. These are consumption rates, not promised improvements in end-to-end completion time. A larger subscription budget can disappear quickly if every task runs in the fastest mode. Speed and billing rules.

All three Pro tiers include a first dot in eligible markets as rollout reaches accounts. Dots are ongoing agents with cloud computers that can work across connected tools and accept continued guidance. Conversations with a dot do not count toward ChatGPT limits; tasks it starts or manages in Work and Codex consume those normal allowances. Value that capability only if it is available to your account and suits your workflow. Introducing dots.

OpenAI overage credits

Plus and Pro users can buy credits to continue eligible Work, Codex, Word, Excel, and PowerPoint activity after included usage runs out. Credits expire twelve months after purchase and are separate from general OpenAI API billing. Purchases are also rolling out to a limited group of Free and Go accounts. Buying credits preserves the existing plan and does not unlock a higher tier's exclusive features. Availability and purchase prices can vary by account, so use the displayed checkout price when comparing overage with an upgrade. Personal-plan credit terms.

Comparing Claude and ChatGPT at the same budget

Free versus Free

Both are worth testing with the same real tasks. Claude Free can demonstrate file creation and Artifacts, while ChatGPT Free offers limited native image generation and Codex. Neither free plan is a reliable substitute for a paid subscription if uninterrupted agent sessions are essential. The useful test is whether each finishes your representative task accurately, and how much correction it needs.

Below 20 USD per month

ChatGPT Go has no directly priced equivalent in Claude's standard individual lineup. At $8, it suits people who need more everyday tool access than Free. Claude Pro's $16.67 effective annual rate is cheaper than paying for it monthly, but requires a $200 commitment. Those are different purchase decisions: a low monthly fee versus a discounted year of a more capable paid package.

Claude Pro versus ChatGPT Plus at 20 USD

This is the most useful comparison for a first paid developer subscription. Claude Pro gives access to Claude Code, Opus, Research, and longer agentic work. ChatGPT Plus provides Codex, Work, broader paid Chat reasoning, and native image generation. Both can analyze files and produce deliverables.

Choose by workflow: test Claude Code and Codex against a representative repository change, including tests and review. For research and writing, compare accuracy, useful citations, revisions needed, and whether the output format fits your process. For images, ChatGPT has an explicit native generation feature. Claude can analyze images and create diagrams and interactive visuals, but Anthropic says it does not generate photos or illustrations in the same way as image-generation tools. Claude image capabilities.

A second $20 subscription can also be rational: two independent tools cost $40 monthly, still far below a $100 upgrade. That buys access to both ecosystems, not a combined quota; whether it is better depends on whether you need diversity of tools or sustained capacity in one agent.

Claude Max 5x versus ChatGPT Pro 100 at 100 USD

Claude's attraction is increased interactive capacity plus $100 in separately usable monthly API credits. OpenAI's attraction is Pro reasoning in Chat and larger Codex and Work allowances. The similarly stated fivefold increases use different baselines and accounting systems, so they do not show equal tokens, equal coding time, or equal completed tasks.

For an API developer, Max's credits can change the economics substantially. For someone who chiefly uses Codex and Astra, Pro 100 may fit better. Compare total subscription and overage costs against time saved on your actual projects.

Claude Max 20x versus ChatGPT Pro 200 at 200 USD

Claude offers twenty times Pro's per-session allowance and $200 in monthly API credits. OpenAI offers its revised Pro 200 allowance, premium Chat reasoning, and Codex and Work capacity. The difference between Claude's advertised 20x and OpenAI's reported 10x cannot be read as Claude offering twice as much useful work: the baselines, models, session rules, and shared budgets differ.

For existing OpenAI subscribers, the October allowance transition also matters. A successful week under a grandfathered allowance is not proof that the post-transition subscription will support the same workload. Test again after the change, or budget around the revised limit now.

The 500 USD tier

Claude's standard individual ladder stops at $200; it has no matching $500 subscription. ChatGPT Pro 500 is worth considering when you need Ultrafast or regularly exhaust lower Pro allowances. At $6,000 for twelve monthly payments, it should earn its place through measured throughput or reduced waiting. For occasional heavy tasks, a lower subscription plus controlled overage may cost less.

Technical differences that affect the purchase

Context windows and uploaded files

Claude's current paid-plan context documentation lists 1 million tokens for Fable 5.1, Opus 5.5, Sonnet 5.5, and Haiku 5.5 in Chat. Context depends on model and surface: for example, Haiku 5.5 is listed at 500,000 tokens in Cowork. Automatic summarization allows long conversations to continue but consumes usage and changes how earlier context is represented. Claude context by model and surface.

ChatGPT's published Chat pricing matrix lists Instant context at 27,000 tokens on Free, 54,000 on Go and Plus, and 128,000 on Pro. Reasoning context is listed as varying on Free, 256,000 on Go and Plus, and 400,000 on Pro. These are total windows, with space used by instructions, tools, memory, and output. A listed window does not grant access to every model, and API or Codex context should not be substituted for Chat limits. ChatGPT context matrix.

Upload capacity is another separate limit. ChatGPT permits documents up to 512 MB and 2 million tokens per file, with approximately 50 MB for spreadsheets; Free has three uploads per day. Projects list five files on Free, twenty-five on Go and Plus, and forty on Pro. Uploading a file larger than the context window does not mean the entire file is simultaneously in working memory. OpenAI's FAQ also says individual-plan document retrieval extracts text rather than embedded images. Upload limits.

Claude's file-creation documentation specifies 30 MB per file for uploads and downloads in that workflow, with larger PDFs potentially processed through its computing environment. For large files, distinguish direct upload, indexed retrieval, and code-based processing before judging either provider by a single headline limit. Claude file workflow.

Research and agent allowances

ChatGPT Deep Research can use the public web, uploaded files, and eligible connected sources, and returns a sourced report. Its current help page directs users to the in-product task counter rather than promising one fixed task total for each plan. Fixed monthly allowances reset thirty days from first use. Deep Research limits and sources.

Claude Research draws on the same general subscription budget as other work; research can use more tokens through repeated searches and analysis. That makes research and coding compete for Claude capacity. On OpenAI, Work and Codex share an agentic allowance, while ordinary Chat premium-model allowances are separately structured. Eligible Word, Excel, and PowerPoint usage can draw from the agentic pool too. Claude Research usage, OpenAI shared allowances.

API access and third-party tools

OpenAI subscription credits extend eligible features; they are not a general API balance. Signing into Codex with an API key invokes separate API billing. Eligible Go, Plus, and Pro subscribers can also opt into sharing their subscription allowance with participating commercial apps through Sign in with ChatGPT. This uses the existing Work and Codex pool, requires separate authorization, and the partner may charge its own fees. Partner usage terms.

Anthropic's 7 October SDK update says the Agent SDK, claude -p, and third-party app usage can still draw from subscription limits; the new Max API credits also cover eligible platform use. Authentication and execution path therefore determine which budget is charged. Do not assume every SDK invocation consumes the newly bundled API balance. Claude Agent SDK billing update.

In interactive Claude Code, logging in with Claude credentials uses the subscription. An ANTHROPIC_API_KEY environment variable can override that path and cause separate API charges. The bundled Max API credit does not fund interactive Code usage even when Code is billed through the Console. Check authentication before drawing conclusions about a subscription's remaining capacity or cost. Claude Code authentication and billing.

Privacy on individual subscriptions

Paying for a personal plan does not itself determine whether content is used for model improvement. OpenAI lets users turn off Improve the model for everyone for new ChatGPT conversations and Codex tasks. Codex has a separate Include environments control; voluntarily submitted feedback can allow the associated conversation to be used even after opting out. OpenAI data controls.

Anthropic also provides model-improvement controls for Free, Pro, and Max, including subscription-backed coding sessions. Incognito chats are not used to improve Claude, although safety-related processing still has exceptions. Check the applicable settings before using proprietary code or sensitive material. Claude consumer privacy.

Announced changes to account for

Three near-term changes deserve attention. Eligible Pro 200 users lose their previous allowance after 29 October. OpenAI also lists GPT-5.5 retirement from Chat, Work, and Codex on 14 October, while API access is unaffected. Finally, custom GPTs are scheduled to retire on 11 December 2026 and migrate to plugins; custom actions do not transfer automatically. Do not build a long-term subscription decision around a feature that already has an announced migration. Pro allowance transition, model retirement notice, custom GPT migration.

The newest announcements also explain conflicting source text. OpenAI's October GPT-6 rollout supersedes older references to Free and Go using GPT-5.6 Luna; its current Pro-tier FAQ supersedes older pause notices. Anthropic's October API-credit terms supersede blanket statements that all personal subscriptions exclude API usage. Anthropic's Pro help page lists high-traffic priority, while its pricing matrix reserves that benefit for Max; do not buy Pro solely for that promise. For a purchase, combine published terms with the features and limits actually shown in your account.

Choosing a plan for your workload

Use one representative week to make the decision. Run the same repository changes, document analyses, and research questions; record whether they finish, correction time, allowance consumed, waiting, and extra charges. Separate model quality from subscription capacity. A cheaper model that completes the task correctly can deliver more value than a premium model used indiscriminately.

Start at $20 unless you already know why you need more. Upgrade for a demonstrated bottleneck, and include API-credit value only when it offsets eligible spending you would otherwise incur. The best plan is the lowest total-cost option that reliably completes your real workload with acceptable quality and delay.