Claude vs ChatGPT in 2026: Pricing and Usage Limits

Pick Claude if your work is long-context coding, agentic automation, or document-heavy analysis, and pick ChatGPT if you want the cheapest capable paid tier and the widest set of consumer features, because both platforms charge $20 per month for their main individual plan and the real difference is in usage limits and model behavior rather than headline price. All figures in this article were verified against Anthropic’s and OpenAI’s official pricing and documentation pages as of August 2026, and both companies change prices and limits frequently enough that you should confirm before you subscribe.

What Claude and ChatGPT Plans Cost as of August 2026

Both vendors run a similar ladder: a free tier, a cheap entry paid tier, a $20 mainstream tier, and high-usage tiers at $100 and $200 per month. The structure is close enough that price alone rarely decides the question.

Anthropic sells Claude Free at $0, Claude Pro at $20 per month billed monthly or $17 per month billed annually (roughly $200 per year), Claude Max 5x at $100 per month, and Claude Max 20x at $200 per month. Team seats run $25 per month billed monthly or $20 per month billed annually for a standard seat, and $125 per month monthly or $100 per month annually for a premium seat that carries roughly five times the usage of a standard seat. Anthropic notes that these are web subscription prices and that mobile app store pricing can differ.

OpenAI sells ChatGPT Free at $0, ChatGPT Go at $8 per month, ChatGPT Plus at $20 per month, and ChatGPT Pro in two tiers at $100 and $200 per month. ChatGPT Business costs $25 per user per month billed monthly or $20 per user per month billed annually, with a two-seat minimum. OpenAI’s $100 Pro tier is documented as five times the usage of Plus, and the $200 tier as twenty times the usage of Plus, which mirrors Anthropic’s Max 5x and Max 20x structure almost exactly.

One notable gap: Anthropic has no direct equivalent to ChatGPT Go at $8 per month. If your budget ceiling is under $20, ChatGPT is currently the only one of the two with a paid option in that range.

How Claude and ChatGPT Usage Limits Actually Work

Neither vendor publishes a simple “you get N messages per month” number, and that is the single most misunderstood part of this comparison. Both meter consumption rather than message count, so a long conversation with large file attachments burns far more of your allowance than a short question.

Anthropic uses a rolling five-hour session window that resets automatically, and paid plans layer a weekly limit on top of it that applies across all models and resets at a fixed time assigned to your account. Usage from claude.ai, Claude Desktop, and Claude Code all draws from the same pool, which matters a great deal if you code with Claude Code and also chat in the browser. Anthropic states plainly that there is no fixed message count because the allowance depends on conversation length, attached files, conversation history, and which model and features you use. Pro, Max, Team, and seat-based Enterprise plans can buy additional usage credits when they run out rather than waiting for the reset.

OpenAI meters in five-hour windows as well, and its published Codex figures are the most concrete usage numbers either vendor discloses. On ChatGPT Plus as of August 2026, OpenAI documents roughly 10 to 100 messages per five-hour window on GPT-5.6 Sol, 25 to 200 on GPT-5.6 Terra, and 250 to 2,000 on GPT-5.6 Luna, with the range reflecting task complexity. The $100 and $200 Pro tiers multiply those by five and twenty respectively. For plain chat, OpenAI documents unlimited everyday text conversation on Free and Go, with separate caps applied to file uploads, image generation, voice, and data analysis.

The practical takeaway is that Claude’s limits are shared across every surface including its coding agent, while ChatGPT’s chat and Codex allowances are described separately. Heavy coders on Claude should expect their browser chat and their terminal work to compete for the same budget.

API Pricing per Million Tokens and Context Window Sizes

If you are building on the API rather than subscribing, the comparison becomes concrete because both vendors publish exact per-token rates. As of August 2026, OpenAI’s flagship is cheaper per token than Anthropic’s, while Anthropic’s mid-tier and OpenAI’s mid-tier are closely matched.

Anthropic’s rates are $10 per million input tokens and $50 per million output tokens for Claude Fable 5, $5 and $25 for Claude Opus 5, $2 and $10 for Claude Sonnet 5, and $1 and $5 for Claude Haiku 4.5. Anthropic has confirmed that the $2 and $10 Sonnet 5 rate, originally announced as introductory pricing, is now the standard price and that a previously scheduled increase to $3 and $15 will not take effect.

OpenAI’s rates are $4 per million input tokens and $20 per million output tokens for GPT-5.6 Sol, $2 and $12 for GPT-5.6 Terra, and $0.20 and $1.20 for GPT-5.6 Luna. OpenAI describes the Sol rate as promotional pricing available at least through November 21, 2026, so treat it as the figure most likely to move.

Two structural details change the real bill more than the sticker rate. Anthropic offers a 50% discount on both input and output through its Batch API and charges cache reads at 10% of the base input rate, with cache writes at 1.25 times base for a five-minute cache or 2 times base for a one-hour cache. Anthropic also includes the full 1,000,000-token context window at standard pricing on Claude 4.6 and later, meaning a 900,000-token request costs the same per token as a 9,000-token one. On the OpenAI side, cached input runs at 10% of the base input rate across the GPT-5.6 family, which is the single largest lever for repetitive agent workloads.

Consumer chat context windows are much smaller than API context windows on both platforms. Anthropic documents up to 1,000,000 tokens for the newest models on paid plans, with 500,000 or 200,000 tokens on other models. OpenAI’s consumer window is substantially smaller than the 1,050,000-token API window and varies by mode and plan, with its fast Instant mode reported at around 32,000 tokens and its Thinking modes considerably higher. OpenAI explicitly cautions that the portion available for your input is smaller than the total window and shifts dynamically based on system instructions, saved memories, and whichever features are active, so treat any single published number as an approximation.

Capability Differences That Show Up in Real Work

Both vendors ship a frontier tier, a balanced tier, and a cheap high-volume tier, and both position them for overlapping jobs. The differences that matter are in how each family is tuned and what surrounding tooling ships with the subscription.

Coding and Agentic Automation

Anthropic explicitly positions Claude Opus 5 for complex agentic coding and enterprise work, and Claude Fable 5 for long-running agents. Claude Code is included with Pro and Max subscriptions rather than sold separately, and Anthropic’s docs recommend Sonnet as the default for the large majority of coding work, reserving Opus for harder problems because it consumes meaningfully more quota. Claude’s models also support adaptive thinking, where reasoning depth adjusts to the task rather than being toggled by hand.

OpenAI ships Codex across Go, Plus, Pro, Business, and Enterprise plans, with the published per-window message allowances noted earlier, and Pro tiers add a research-preview coding model on top of the standard lineup. If you want a documented, countable coding allowance before you pay, OpenAI currently discloses more than Anthropic does.

Writing and Everyday Chat

For prose, both flagship families handle long-form drafting well, and the deciding factor is usually cost per output token, since writing is output-heavy. GPT-5.6 Sol at $20 per million output tokens is meaningfully cheaper than Claude Opus 5 at $25 and less than half of Claude Fable 5 at $50. For high-volume, low-stakes generation, GPT-5.6 Luna at $1.20 per million output tokens is the cheapest option in this comparison by a wide margin. On subscriptions rather than API, the $20 tiers are equivalent in price and the choice comes down to which model’s voice you prefer, which is genuinely a matter of taste.

Research and Long-Document Analysis

This is where the context window gap matters most. On paid Claude plans, the newest models expose up to a 1,000,000-token window in the consumer product, which is enough to hold a large codebase or a stack of contracts in a single conversation. OpenAI’s consumer context is documented at a much smaller size, so heavy document work in ChatGPT typically requires chunking or file-search retrieval rather than pasting everything in.

Knowledge cutoffs also differ. Anthropic documents a May 2026 training data cutoff for Claude Opus 5 and January 2026 for Claude Fable 5 and Claude Sonnet 5. OpenAI documents a February 16, 2026 knowledge cutoff for the GPT-5.6 family. For questions about the recent past, Claude Opus 5 has the fresher baseline, though both platforms rely on live web search for anything current.

Which One to Pick for Your Workflow

Skip the feature checklists and match your actual usage pattern to a tier. As of August 2026, these are the clean decision rules.

• Pick ChatGPT Go at $8 per month if you want a paid tier under $20 and mostly need everyday chat with occasional file uploads and image generation. Anthropic has nothing at this price.
• Pick Claude Pro at $20 per month if you write code daily and want Claude Code included, or if you regularly work with documents large enough to need a very large context window in a chat interface.
• Pick ChatGPT Plus at $20 per month if you want the broadest consumer feature surface and a documented, predictable coding allowance, or if your work is output-heavy writing where per-token cost favors the GPT-5.6 family.
• Pick Claude Max 5x or ChatGPT Pro at $100 per month if you hit the $20 tier’s limits several times a week. Both deliver five times the mainstream tier, so pick whichever platform you already prefer rather than shopping on price.
• Pick the $200 tier on either platform if you run an agent or coding workflow most of the working day. Below that intensity, $200 per month is very difficult to justify against API pay-as-you-go.
• Pick the API instead of a subscription if your usage is bursty or programmatic. A workload of 10 million input and 2 million output tokens per month costs $44 on GPT-5.6 Terra and $40 on Claude Sonnet 5, well under a single $100 subscription, and Anthropic’s 50% Batch API discount and 10% cache reads can cut that further.
• Pick both if you are a professional user with a budget over $40 per month. Two $20 subscriptions cost the same as one $40 upgrade and give you a second opinion on hard problems, which is often worth more than doubled limits on one platform.

A word of caution on the arithmetic: because neither vendor publishes message counts for general chat, the only reliable way to size your plan is to run the $20 tier for a full month and check whether you are hitting limits. Upgrading before you have that data usually buys headroom you do not use.

Claude vs ChatGPT Pricing FAQ

Is Claude or ChatGPT cheaper in 2026?

They cost the same at the mainstream tier. Both Claude Pro and ChatGPT Plus are $20 per month as of August 2026, and both vendors offer $100 and $200 high-usage tiers. ChatGPT is cheaper at the entry level because ChatGPT Go costs $8 per month and Anthropic has no equivalent plan. On the API, OpenAI’s flagship GPT-5.6 Sol at $4 per million input and $20 per million output tokens undercuts Claude Opus 5 at $5 and $25, though OpenAI describes the Sol rate as promotional through at least November 21, 2026. Claude Sonnet 5 and GPT-5.6 Terra are effectively tied at $2 per million input tokens.

How many messages do you get on Claude Pro and ChatGPT Plus?

Neither vendor publishes a fixed message count for general chat, because both meter total consumption rather than message count. Anthropic resets usage on a rolling five-hour session window with an additional weekly limit on paid plans, and states that your actual allowance depends on message length, attached files, conversation history, and the model in use. OpenAI publishes concrete figures only for Codex, where ChatGPT Plus is documented at roughly 10 to 100 messages per five-hour window on GPT-5.6 Sol, 25 to 200 on GPT-5.6 Terra, and 250 to 2,000 on GPT-5.6 Luna as of August 2026. For plain text chat, OpenAI documents unlimited everyday conversation on Free and Go with separate caps on files, images, and voice.

Do the free tiers of Claude and ChatGPT give you the top models?

Both free tiers give you access to current models with substantially reduced capacity, not a locked-down older model. Anthropic’s pricing page lists the Fable, Opus, Sonnet, and Haiku families as available on the Free plan at 50% of the weekly limits that paid plans receive, and Claude Pro is documented as offering at least five times the usage per session that Free does. OpenAI has made GPT-5.6 Luna the default model for Free and Go users, with unlimited everyday text chat and separate limits on file uploads, image generation, voice, and data analysis. In both cases the free tier is a genuine evaluation environment, and the paid tier buys capacity rather than unlocking a different class of model.

Related reading

• Claude Sonnet 5’s 1M-Token Context Window Explained
• DeepSeek Peak-Hour Pricing: What Changes on August 16, 2026

More on this topic: all AI Tools articles · full article index