Free Tool

Claude Model Selector: Opus, Sonnet or Haiku?

ClaudeAIHub.

Not sure which Claude model to use? Answer five questions and get a starting recommendation. The right model depends on your task complexity, cost sensitivity, and whether you’re using the app or the API.

Free Browser Tool

Answer questions about your task type, output length, speed requirements, and budget sensitivity. The selector maps your answers to the Claude model best suited to your use case — Claude Opus 5.5, Sonnet 5.5, Haiku 4.5, or Fable 5.1.

What is your main task?

What matters most to you?

How large are your inputs?

How long are the outputs you need?

How will you use Claude?

Claude Mythos 5.1 is a restricted, trusted-access model and is not included in these recommendations. Model availability and capabilities change — always check Anthropic's model docs before choosing. ClaudeAIHub.

Claude Models Explained

Claude Fable 5.1

Anthropic’s top-tier model, launched September 1, 2026 as the successor to Fable 5. Choose it for demanding reasoning and long-horizon agentic work — or when your evals on Opus 5.5 at high effort still fall short. Priced at $10/$50 per MTok via the API, with cache reads at $0.25/MTok. See the Claude Fable 5.1 guide for specs, access, and safeguards.

Claude Opus 5.5

Anthropic’s recommended default, released September 22, 2026. Best for agentic coding, research, knowledge work, and multi-step planning. It outscores Fable 5.1 on Anthropic’s published benchmarks at $4 / $20 per MTok, and it’s the default model in Claude Code on every paid plan.

Claude Sonnet 5.5

The best combination of speed and intelligence, released September 28, 2026. Handles most professional writing, research, coding, and business tasks, including sustained agentic tool use, at $2 / $10 per MTok — and comes close to Opus 5.5 on many benchmarks while responding faster.

Claude Haiku 4.5

The fastest model in the current lineup. Designed for high-volume, latency-sensitive API workloads where cost and speed matter more than maximum capability. Good for simple tasks, classification, fast Q&A, and processing large batches. Available via API.

Model Comparison

Feature	Claude Opus 5.5	Claude Sonnet 5.5	Claude Haiku 4.5
Best for	Recommended default: agentic coding, knowledge work	Balance of speed and intelligence	Speed, cost efficiency, scale
Context window	1M tokens	1M tokens	200k tokens
Max output	128k tokens	128k tokens	64k tokens
Input price	$4/MTok	$2/MTok	$1/MTok
Output price	$20/MTok	$10/MTok	$5/MTok
Extended thinking	No (adaptive, always on)	No (adaptive; `between_tools` lowest setting)	Yes
Adaptive thinking	Yes	Yes	No
Relative speed	Moderate	Fast	Fastest

Claude Fable 5.1 sits above this table: $10/MTok input, $50/MTok output, $0.25/MTok cache reads, 1M token context, 128k max output, adaptive thinking always on. Claude Mythos 5.1 shares Fable 5.1’s specs but is restricted to Anthropic’s Project Glasswing trusted-access program and is not a normal selector option. Claude Sonnet 5.5 keeps the $2/$10 per MTok price made permanent for Sonnet 5 in August 2026. Verify current pricing and availability at platform.claude.com/docs/en/about-claude/models/overview.

App Users vs API Users

Claude app users (claude.ai, desktop, mobile): The model you access depends on your subscription plan. The Free plan provides access to limited model usage. Pro gives access to Sonnet. Max and higher tiers give access to Opus. You don’t choose a model ID directly — the app selects based on your plan and the task.

API users: You choose the model ID explicitly in each API request. You can use any available model regardless of what app plan you have. API billing is separate from app plans — you pay per token.

When to Try a Different Model

Output quality is too low: Try the next tier up. Moving from Haiku to Sonnet often significantly improves structured or reasoning-heavy tasks.
Cost is too high: Try the next tier down. Many tasks that feel like they need Opus work just as well on Sonnet.
Latency is too high: Try Haiku for time-sensitive applications where response speed matters more than depth.
Context window exceeded: Move from Haiku (200k) to Sonnet or Opus (1M) for long documents.