Skip to main content
Claude comes in three lines — Opus, Sonnet and Haiku — and choosing between them usually comes down to “how smart” versus “how fast and cheap.” This page helps you pick a model for the task and understand what Claude can do. For the technical side of connecting, see the Quickstart; this page is about the models themselves.

​
How the lines differ

LineStrong atSpeed / costReach for it when
Claude OpusThe hardest reasoning and large codeSlower, pricierAgents, multi-step tasks, heavy refactors
Claude SonnetAll-rounder: code, analysis, long contextMidEveryday work — the best balance
Claude HaikuSimple tasks at high volumeFaster, cheaperClassification, short replies, bulk processing
Not sure? Start with Sonnet. It covers most work; move up to Opus when Sonnet stops keeping up, and down to Haiku when speed and cost matter more than peak quality.

​
What Claude can do

  • Long context — a large window that fits whole codebases and documents.
  • Vision — you can pass images as input: screenshots, diagrams, scans. See image input for the request format.
  • Function calling (tool use) — the model can return structured tool calls; agents are built on this. See function calling for a full example.
  • Thinking mode — Sonnet and Opus have an extended “reasoning before answering” mode, useful in math, logic and hard code.
For long answers, you can also stream the response token by token instead of waiting for the full reply.
The model field takes an ID, e.g. claude-opus-4-8 for the top model. Exact IDs, versions and per-token prices are on the Pricing page at www.ruapi.ai. Names match Anthropic’s official ones, and new versions appear in the catalog automatically.

​
Picking by task

  • Writing and refactoring code → Sonnet by default, Opus for the gnarliest tasks.
  • Building an agent or tool pipeline → Opus or Sonnet (you want reliable tool use).
  • Analyzing long documents → Sonnet with long context.
  • Classification, tagging, short replies at scale → Haiku, for speed and cost.
  • Images as input → Sonnet and Opus with vision.

​
How to connect

Claude works over two protocols — Anthropic (Claude Code, Anthropic SDK) and OpenAI-compatible (Cursor, Cline, OpenAI SDK). Step-by-step base_url setup and a first request are in the Quickstart; full Claude Code config is in its guide.

​
FAQ

Sonnet is the everyday pick and noticeably cheaper. Use Opus selectively, when Sonnet can’t carry the complexity: long reasoning chains, large refactors, demanding agents.
Large — Sonnet and Opus fit whole projects and long documents. The exact limit depends on the version and matches Anthropic’s official one.
Yes, the vision models (Sonnet, Opus) take images as input — screenshots, diagrams, scans.
Yes, Sonnet and Opus have an extended thinking mode — the model “thinks through” the task before answering, which helps with math, logic and hard code.
Yes. Claude can return structured tool calls — the basis for agents and integrations with external APIs.
On the main site’s Pricing page. Names match Anthropic’s official ones, so you use the same IDs in code.

​
Next

Connect and first request

base_url, key, Python and curl code.

Claude Code: full setup

Environment variables and common errors.