Stack Picker
a developer-grade decision engine
Back to the picker
AI / LLM

Anthropic (Claude)

Claude — strongest coding and reasoning models.

Official site
Monthly cost
$5–$50 / mo
Usage-based ~$5–$50/mo light usage
Popularity
5/5
LLM knowledge
4/5
Difficulty
Easy
#ai-native

What Anthropic (Claude) is good at

Strengths
  • +Best at coding
  • +Long context
  • +Tool use works well
Tradeoffs
  • Fewer modalities than GPT
  • Rate limits at launch

Coding-agent prompt

You're calling the Anthropic Claude API. Follow these rules:

- Use the official SDK: `@anthropic-ai/sdk`.
- Start with `claude-sonnet-4-6` for general tasks; `claude-opus-4-7` for harder reasoning; `claude-haiku-4-5` for latency/cost.
- Use `messages.create` with a `system` prompt + `messages` array. Keep system prompts short and strict.
- Prompt cache with `cache_control: { type: "ephemeral" }` on large, repeated context (docs, system prompts).
- Tool use via `tools`; validate tool inputs with zod before acting on them.
- Stream for anything the user is watching (`stream: true`).
- Keep `ANTHROPIC_API_KEY` server-side; never ship it to the client.

Beginner's guide to Anthropic (Claude)

In one line: Use Claude, an AI model, in your app via an API.

Anthropic's API lets you send text to Claude and get text back. Great for AI features — summarizing, answering questions, writing code. You pay per million tokens (roughly, per million characters) used.

Try it in your terminal
  • npm install @anthropic-ai/sdk

    Install the official SDK.

Browse all categories