AI / Claude Models Basics Interview Questions
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.
| Model | Input (per MTok) | Output (per MTok) |
|---|---|---|
| Claude Fable 5 | $10 | $50 |
| Claude Opus 4.8 | $5 | $25 |
| Claude Sonnet 5 | $3 (intro: $2 until Aug 31 2026) | $15 (intro: $10) |
| Claude Haiku 4.5 | $1 | $5 |
Cost-saving features:
- Prompt caching — reuse of cached prompt prefixes is billed at a significant discount (cache writes are more expensive; cache reads are cheaper than standard input)
- Message Batches API — async batch processing at roughly 50% of standard pricing, ideal for large-scale, non-real-time workloads
Cloud platform pricing (Amazon Bedrock, Google Cloud) may differ from direct API pricing. See the Pricing page for full details including per-region variations.
Which Claude model costs $1 per million input tokens?
What is the main benefit of using the Claude Message Batches API?
More Related questions...
