Context window and limits
The 1M-token window, output limits, text-only input and other limits.
- Can I upload files to SuperCosmos?The model reads text. Products handle files their own way: Line CLI reads your project files; the Model API takes text in messages.FAQ
- Can SuperCosmos read images, PDFs or audio?SuperCosmos reads text only. Extract text from PDFs first; image input isn't supported.FAQ
- Can SuperCosmos read my whole codebase at once?If it fits in 1M tokens, yes. In practice Line CLI reads the files a task needs, which is faster and cheaper.FAQ
- Can SuperCosmos tell me today's news or prices?Not by itself. Without a tool it can't see live data, and it's built to say so instead of making numbers up.FAQ
- Does the 1M-token window include SuperCosmos's answer?Yes. The window holds everything in the request (instructions, history, documents, tool results) plus the reply being written.FAQ
- Does using more of the context window cost more?Yes. You pay for every input token, and prompts over 262K tokens bill their whole input at 1.5×.FAQ
- How do I get good answers from very long inputs?Put the question after the material, point to the sections that matter, and ask it to quote what it relies on.How-to
- How do I see how much of the context window I'm using in Line CLI?Run /context: it shows how much of the model's window is in use, broken down. Line also shows a live context meter.How-to
- How long can a SuperCosmos answer be?Up to 32,768 output tokens on Flash and Max and 49,152 on Heavy, reasoning included.Reference
- How many messages can a SuperCosmos conversation have?There's no fixed message count for the model; the 1M-token window is the limit. Products add their own limits (the Model API: 800 per request).FAQ
- What are SuperCosmos's limits in each product?The model's 1M window is shared; each product adds its own limits (Model API 3 MB / 800 messages, Agent API input 20,000 characters).Reference
- What happens when a Line CLI session fills the context window?Line CLI auto-compacts at about 80% of the 1M-token window, summarising older steps so the task can continue.FAQ
- What is a context window, in simple words?The context window is how much text the model can look at in one go: your question, the chat so far, and its answer.Explainer
- What is SuperCosmos's context window?1M tokens (1,048,576), shared by Flash, Max and Heavy. Prompts over 262K tokens bill input at 1.5×.Reference
- What is SuperCosmos's knowledge cutoff?Clatter Labs doesn't publish a cutoff date. For recent facts, give the information in your prompt or use a tool.FAQ
- Why does SuperCosmos forget the start of a very long chat?If a conversation outgrows the window, earlier parts are summarised or dropped by the product (or your app).Troubleshooting
Still need help?
Tried the steps above and still stuck? Email us. We usually reply within the first hour, and always within 24 hours. A full support team is with you.
Infrastructure questions, or permanent deletion of your data
infrastructure@clatterlabs.xyzFor deleting your whole account or the data of one service.