Claude’s context window is 1 million tokens on every current model (Claude Fable 5.1, Opus 5.5, Sonnet 5.5 and Haiku 5.5), which Anthropic says is roughly 555,000 words. But will your document fit? Paste it into this free checker, or enter a word, page or token count, and see how much of each Claude, GPT and Gemini model’s context window it fills.
It estimates tokens with each maker’s own rule of thumb, leaves room for the answer, and runs entirely in your browser. Nothing you paste is uploaded.
Free checker
Will it fit? Check any AI context window
Paste your text, or enter a word, page or token count. You see right away whether it fits the context window of each major Claude, GPT and Gemini model, using the numbers each maker publishes.
Where you use it
Show
Export the full comparison table
Unlock CSV download and copy-as-Markdown for this table, with every model ID, window, input and output limit, your estimate and the source link, ready for a doc, a ticket or a spreadsheet.
Unlocked: the export buttons are just below. You are also on the weekly HOW TO // AI newsletter.
Free. You also get the HOW TO // AI newsletter once a week. Unsubscribe any time.
Need an exact-ish token count or a cost? Try the Claude token counter or the AI token calculator.
How we estimate: Claude models from Opus 4.7 on, about 2.5 characters or 0.555 words per token (Anthropic's models overview, which also notes about 30% more tokens than older Claude models in its token counting docs); older Claude models, 0.75 words per token; GPT, 4 characters or 0.75 words per token (OpenAI); Gemini, 4 characters or about 0.7 words per token (Google). Pasted text averages the character and word estimates. Context windows: Anthropic, Claude Help Center (apps), OpenAI model pages, Google Gemini models and Gemini plans. Checked October 18, 2026. Estimates only: count exactly with each maker's tokenizer before you rely on a tight fit.
How to use the context window checker
- Paste your text, or switch to Words, Pages or Tokens and type a number.
- Set how much room to leave for the answer. A long reply can run to several thousand tokens.
- Switch between the API and the chat apps to see paid-plan limits.
- Read the verdict for each model: Fits, Tight or Too big. If something is too big, the tips under the table show what to do.
Context window sizes: Claude, GPT and Gemini
| Model | Context window | Notes |
|---|---|---|
| Claude Fable 5.1, Opus 5.5, Sonnet 5.5, Haiku 5.5 | 1,000,000 tokens | Output up to 128,000. Input and output share the window. |
| Claude Sonnet 4.6 (legacy) | 1,000,000 tokens | 500,000 when chatting on paid Claude plans |
| Claude Sonnet 4.5 (deprecated), Claude Haiku 4.5 | 200,000 tokens | Haiku 4.5 output up to 64,000 |
| GPT-6 Astra, GPT-6.1 Sol, GPT-6 Luna | 1,050,000 tokens | Input up to 922,000; output up to 128,000 |
| GPT-5.5 | 1,050,000 tokens | Output up to 128,000 |
| GPT-5 | 400,000 tokens | Input up to 272,000; output up to 128,000 |
| GPT-4.1 | 1,047,576 tokens | Output up to 32,768 |
| GPT-4o | 128,000 tokens | Output up to 16,384 |
| Gemini 3.8 Flash, 3.1 Pro (preview), 3.5 Flash-Lite, 2.5 Pro, 2.5 Flash | 1,048,576 input tokens | Output up to 65,536, counted separately |
Looking for Gemini 3 Pro? Google lists Gemini 3 Pro Preview as shut down. Gemini 3.1 Pro (preview) is the current Pro model, with the same 1,048,576-token input limit.
Claude’s context window in the apps
When you chat with Claude on a paid plan (Pro, Max, Team or Enterprise), the current models also get a 1M-token window. Claude Sonnet 4.6 gets 500K, and other older models 200K. In Claude Code, every current model has a 1M-token window.
Long chats don’t simply stop at the limit: on paid plans with code execution turned on, Claude summarizes earlier messages automatically as a conversation nears the limit, so it can keep going. Those longer conversations use more of your usage limit.
Google says its Google AI Pro plan includes a 1M-token context window in the Gemini app. OpenAI doesn’t publish one context window per ChatGPT plan that we could verify, so the checker shows OpenAI’s models through the API only.
Why Claude counts more tokens now
Claude Opus 4.7 and later use a newer tokenizer that turns the same text into about 30% more tokens. So the same 1M-token window holds about 555,000 words on current Claude models, compared with about 750,000 on Claude models before Opus 4.7. That’s why the checker shows a higher token count for current Claude models than for GPT or older Claude models. Anthropic’s advice: recount your prompts on the model you plan to use rather than reusing old counts.
What counts, and what to do when it doesn’t fit
- Everything in the request counts. For Claude that’s the system prompt, every message, documents, images, tool definitions and the reply itself, thinking included. OpenAI’s window likewise covers input, output and reasoning tokens.
- Fitting isn’t the same as working well. Anthropic notes that accuracy and recall drop as the token count grows, so send what the task needs, not everything you have.
- Too big? Send only the relevant sections, split the document into parts, or summarize each part first and work from the summaries.
- Building on the API? For long-running conversations, Anthropic recommends server-side compaction. Before a big request, count it exactly with the maker’s token counting tool.
Related on HOW TO // AI
- Claude token counter: count tokens and cost for current Claude models.
- AI token calculator: tokens, words and cost for any model.
- Claude pricing: every plan, with usage limits.
- AI chatbot pricing compared: ChatGPT, Claude, Gemini and more side by side.
Checked against Anthropic, OpenAI and Google model documentation, the Claude Help Center and Google’s Gemini plans page on October 18, 2026.
Frequently asked questions
What is Claude's context window?
1 million tokens on every current Claude model: Fable 5.1, Opus 5.5, Sonnet 5.5 and Haiku 5.5. Anthropic says that is roughly 555,000 words. Older models such as Claude Sonnet 4.5 and Haiku 4.5 have 200,000 tokens.
What is a context window?
It is how much text a model can work with at once, its working memory. It holds your prompt, the conversation so far, any documents and the reply, and it is measured in tokens.
How many pages fit in 1 million tokens?
On current Claude models, Anthropic says 1 million tokens is roughly 555,000 words, or about 1,100 single-spaced pages at 500 words a page. With OpenAI's rule of thumb of 0.75 words per token, it is about 750,000 words.
What is GPT-5's context window?
OpenAI lists GPT-5 at 400,000 tokens, with input up to 272,000 and output up to 128,000. Its newer GPT-6 models have 1,050,000 tokens, with input up to 922,000.
What happens when a chat goes past the context window?
On paid Claude plans with code execution turned on, Claude summarizes earlier messages automatically so the chat can continue. In the API you manage it yourself, and Anthropic recommends server-side compaction for long conversations.