AI / Claude Models Basics Interview Questions
1. What is Claude and who makes it?
Claude is a family of state-of-the-art large language models (LLMs) built by Anthropic , an AI safety company founded in 2021. Claude excels at language, reasoning, analysis, coding, mathematics, and creative writing, and is designed with a strong focus on being helpful, harmless, and honest. Ant...
2. What are the current Claude model families and what is each one optimised for?
Claude models are organised into three tiers — Opus , Sonnet , and Haiku — representing a capability-speed-cost spectrum. As of mid-2026 the current flagship generation is Claude 4/5, alongside the newly released Claude Fable 5. Current Claude model tiers Tier Optimised for Example model Opus Max...
3. What are the API model IDs for the current Claude models?
When making API calls you must specify the exact model ID string. Model IDs starting with the Claude 4.6 generation use a dateless format that is still a pinned snapshot (not an evergreen pointer). Earlier models used a dated format like claude-haiku-4-5-20251001 . Current model IDs (Claude API) ...
4. What is a context window and what are the context window sizes for current Claude models?
A context window is the total number of tokens (words, punctuation, code, etc.) that a model can process in a single request — encompassing the system prompt, all conversation history, tool definitions, and the model's own output. If you exceed the context window, older content must be removed or...
5. What are the pricing tiers for current Claude models and how is pricing calculated?
Claude API pricing is charged per million tokens (MTok) — counting both input tokens (your prompt, system prompt, conversation history) and output tokens (Claude's response). Prices differ by model and reflect the capability-cost trade-off. Current Claude API pricing (per million tokens) Model In...
6. 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. Modality support across current models Capability Supported? Notes Text input Yes All models Image input (vision) Yes All model...
7. What is extended thinking and how does it differ from adaptive thinking in Claude?
Both features enable Claude to reason more carefully before answering, but they work differently and are available on different models. Extended thinking vs Adaptive thinking Feature Extended Thinking Adaptive Thinking What it does Claude produces explicit blocks showing its reasoning steps, visi...
8. What platforms and cloud providers is Claude available on?
Claude is available through multiple channels, each with its own model IDs, endpoint behaviour, and pricing structure. Claude availability by platform Platform Best for Model ID format Claude API (direct) Developers building directly with Anthropic claude-opus-4-8, claude-sonnet-5, etc. claude.ai...
9. What is the knowledge cutoff for current Claude models?
Claude models are trained on data up to a specific date (the training data cutoff ) and have the most reliable knowledge through a slightly earlier date (the reliable knowledge cutoff ). Claude does not have access to real-time internet data during a conversation unless given a search tool. Knowl...
10. What is the Claude model lifecycle — what do 'Active', 'Legacy', and 'Deprecated' mean?
Anthropic uses a defined set of lifecycle statuses for Claude models. Understanding these helps teams plan migration timelines and avoid unexpected outages. Model lifecycle statuses Status Meaning Action required? Active Fully supported and recommended for new development No — this is the ideal s...
11. What is Claude Fable 5 and what makes it different from Claude Opus 4.8?
Claude Fable 5 ( claude-fable-5 ) is Anthropic's most capable widely-released model as of mid-2026, positioned above the Opus tier. It is designed for long-running agents, frontier intelligence tasks, and complex enterprise work. Claude Fable 5 vs Claude Opus 4.8 Feature Claude Fable 5 Claude Opu...
12. What is Claude Mythos 5 and how does it differ from Claude Fable 5?
Claude Mythos 5 ( claude-mythos-5 ) is a variant of Claude Fable 5 that is offered separately for defensive cybersecurity workflows as part of Anthropic's Project Glasswing . It is not publicly available — access is invitation-only with no self-serve sign-up. Claude Fable 5 vs Claude Mythos 5 Fea...
13. What is Claude Haiku 4.5 and what are its key characteristics?
Claude Haiku 4.5 ( claude-haiku-4-5-20251001 , alias claude-haiku-4-5 ) is Anthropic's fastest and most cost-efficient model in the current generation. It is described as achieving near-frontier performance on coding, computer use, and agent tasks while being optimised for speed and low latency. ...
14. What is prompt caching and how does it reduce costs when using Claude?
Prompt caching allows Anthropic to store a copy of a prompt prefix (such as a long system prompt, documentation, or conversation history) so that subsequent requests reusing that prefix are billed at a much lower rate than re-sending it fresh each time. Prompt caching pricing structure Token type...
15. What is the Messages Batches API and when should you use it?
The Message Batches API allows you to submit a large number of Claude API requests asynchronously in a single batch, receiving results once all requests are processed. It is designed for large-scale, non-time-sensitive workloads. Messages Batches API vs standard API Feature Standard Messages API ...
16. What is tool use (function calling) in Claude and which models support it?
Tool use (also called function calling) allows Claude to request the execution of external functions and incorporate their results into its responses. You define a set of tools with names, descriptions, and input schemas; Claude decides when to call them and how to structure the arguments. # Defi...
17. What is computer use in Claude and which models support it?
Computer use is a set of built-in tools that allow Claude to interact directly with computers — taking screenshots, moving the mouse, clicking, typing, and running bash commands. It is designed for agentic automation tasks where Claude operates a full computer desktop or terminal environment. Com...
18. What are the different claude.ai plans and what does each include?
claude.ai offers multiple subscription tiers designed for individuals, teams, and enterprises. Each tier provides different levels of usage, features, and access to Claude models. claude.ai plans overview Plan Who it's for Key features Free Individual — casual use Access to Claude; usage limits; ...
19. What is the effort parameter in Claude and which models support it?
The effort parameter allows you to trade intelligence for latency and cost within a single model — rather than switching to a different model. It is available on recent Opus and Sonnet models. Effort parameter levels Level Behaviour Use case low Fastest, least compute — lighter reasoning Simple t...
20. What is streaming in Claude API responses and how do you use it?
Streaming allows you to receive Claude's response token by token as it is generated, rather than waiting for the complete response. This dramatically reduces the time to first token and creates a more responsive user experience for chat applications. # Streaming with the Python SDK with client . ...
21. What is the system prompt in Claude and how does it affect model behaviour?
The system prompt is an optional instruction block passed at the start of a conversation that sets Claude's persona, context, constraints, and behavioural guidelines before the first user message. It is processed before any human turn and shapes how Claude responds throughout the conversation. # ...
22. What is zero data retention (ZDR) and which Claude models support it?
Zero data retention (ZDR) is a data handling agreement where Anthropic does not store API inputs or outputs after a response is returned. This is important for organisations with strict data privacy requirements (healthcare, legal, finance) where conversation data must not persist on Anthropic's ...
23. What is Claude's approach to safety and what are Constitutional AI principles?
Anthropic builds Claude with a strong emphasis on AI safety — designing the model to be helpful, honest, and to avoid causing harm. The primary training technique underpinning Claude's values is Constitutional AI (CAI) . Constitutional AI works by training the model against a set of written princ...
24. What is the difference between an operator and a user in Claude's design?
Anthropic distinguishes between two types of principals who interact with Claude: operators and users . This distinction matters because it determines the level of trust Claude extends to instructions and how it resolves conflicting requests. Operator vs User Aspect Operator User Who they are Com...
25. What is Claude's context window and how are tokens counted?
Claude's context window is the total number of tokens it can process in a single API request. Tokens are the fundamental unit of text that Claude processes — roughly 3-4 characters per token for English, or about 75% of a word on average. Token counting rules of thumb Content type Approximate tok...
26. What are Claude's rate limits and how are they structured?
Claude API rate limits prevent overload and ensure fair access. They are applied at three levels: requests per minute (RPM), tokens per minute (TPM), and tokens per day (TPD). Limits vary by model and by API usage tier. Rate limit dimensions Limit type What it restricts Requests per minute (RPM) ...
27. What is Claude's approach to harmful content — what will and won't it do?
Claude has hardcoded behaviours (absolute limits that cannot be changed by any instruction) and softcoded defaults (behaviours that operators or users can adjust within permitted ranges). Understanding this distinction helps developers build applications that work well within Claude's guidelines....
28. What is Claude's max_tokens parameter and how does it relate to the context window?
The max_tokens parameter sets the maximum number of output tokens Claude will generate in a single response. It is a hard cap — Claude will stop generating once it reaches this limit, potentially truncating its response mid-sentence. # max_tokens is required in the Messages API response = client ...
29. What is the temperature parameter in Claude and how does it affect responses?
The temperature parameter controls the randomness of Claude's output. Higher temperatures produce more varied, creative responses; lower temperatures produce more focused, deterministic responses. Temperature settings Value Behaviour Best for 0 Deterministic — same input almost always gives same ...
30. What are Claude's multimodal capabilities — how does it process images and documents?
Claude's vision capabilities allow it to analyse and reason about images, PDFs, and screenshots alongside text. This makes it useful for document analysis, UI debugging, chart interpretation, and more. import anthropic , base64 client = anthropic . Anthropic() # Option 1: URL-based image (Claude ...
31. What are the claude.ai plans and what models does each tier include access to?
claude.ai offers consumer and business plans, each with different model access and usage limits. The model you can use in the chat interface depends on your subscription tier. claude.ai model access by plan Plan Models available Usage limits Free Claude (typically Haiku or Sonnet) Limited — daily...
32. What is multi-turn conversation handling in Claude and how do you implement it?
Claude's Messages API is stateless — each API call is independent and Claude has no memory of previous calls unless you include the conversation history explicitly. Multi-turn conversation is implemented by appending each exchange to the messages array. # Building a multi-turn conversation manual...
33. What are the different stop_reason values in Claude API responses?
Every Claude API response includes a stop_reason field indicating why Claude stopped generating. Understanding stop reasons is essential for building robust applications — especially for tool use and handling truncated responses. stop_reason values Value Meaning Action required? end_turn Claude n...
34. What is Claude's approach to honesty and what does it mean for Claude to be non-deceptive?
Honesty is a central Claude value. Anthropic designs Claude to have a cluster of honesty-related properties that go beyond simply not lying — covering how Claude represents uncertainty, its own nature, and its limitations. Claude's honesty properties Property What it means Truthful Only sincerely...
35. What is Claude Code and how does it differ from using Claude directly via the API?
Claude Code is Anthropic's agentic coding tool — a command-line interface (CLI) and SDK that allows Claude to work autonomously on coding tasks in your terminal, with direct access to your file system, git, and development tools. Claude Code vs Claude API Feature Claude Code Claude API (direct) I...
36. What are the Anthropic SDKs and what languages are officially supported?
Anthropic provides official SDKs that wrap the Claude API, handling authentication, request formatting, response parsing, automatic retries, and streaming. Using an SDK is strongly recommended over direct HTTP calls. Official Anthropic SDKs Language Package Install command Python anthropic pip in...
37. What is Anthropic's policy on model deprecation and how should developers prepare?
Anthropic has a formal model lifecycle and deprecation policy to help developers plan migrations without unexpected disruptions. Knowing this policy helps you build more resilient applications. Anthropic deprecation policy key points Policy Detail Minimum notice At least 60 days before any public...
38. What are the key differences between Claude 4 and earlier Claude 3 generation models?
Claude 4 (and the Claude 4/5 generation more broadly) represents significant advances over Claude 3 across capability, context, and new features. Understanding what changed helps teams make informed migration decisions. Claude 3 vs Claude 4+ generation comparison Feature Claude 3 generation Claud...
