AI / Claude Models Basics Interview Questions
What input and output modalities do current Claude models support?
All current Claude models share a common set of supported modalities for input and output, with no difference between Opus, Sonnet, and Haiku tiers on core capabilities.
| Capability | Supported? | Notes |
|---|---|---|
| Text input | Yes | All models |
| Image input (vision) | Yes | All models — up to 600 images per request (100 for 200k models) |
| PDF input | Yes | Treated similarly to images for token budgeting |
| Text output | Yes | All models |
| Multilingual | Yes | Strong performance across major languages |
| Tool use / function calling | Yes | All models |
| Extended thinking | Haiku 4.5 only | Explicit thinking steps visible in output |
| Adaptive thinking | Opus and Sonnet (not Haiku 4.5) | Always-on for Fable 5 |
| Audio input | No | Not currently supported |
| Video input | No | Use frame extraction for video analysis |
Vision notes: Claude models can analyse images, charts, screenshots, UI elements, and document scans. For video analysis, the recommended approach is to extract frames and send them as a series of images. Claude Opus 4.5 and 4.6 showed improved vision capabilities — especially for multi-image tasks and computer use.
Which input modality is NOT currently supported by any Claude model?
Which Claude model uniquely supports Extended Thinking (explicit step-by-step reasoning visible in output)?
More Related questions...
