LaunchpadHQ
All model families

Anthropic

Claude models

Version comparison

Last verified Oct 14, 2026. Figures that could not be confirmed on an official vendor page are marked unverified — we do not guess.

Full Claude comparison

Which one should I use?

  • On claude.ai, start with Sonnet for everyday writing, coding, and analysis. It is the default on the free plan.
  • Use Haiku when you want a fast, cheap answer and the job is simple (summaries, sorting, short rewrites).
  • Use Opus when the task is long and you need it to keep working without you steering every step. That needs a paid plan.
  • Use Fable only for the largest, longest jobs. Anthropic positions it above Opus; it uses more of your usage limit.
  • API developers: Anthropic currently recommends Opus 5.5 as the default, Haiku 5.5 for high-volume cheap work, and Fable 5.1 when Opus is not enough.

Claude Opus 5.5

Current
Context
1M
Max output
128K
API input / 1M
$4 / 1M
API output / 1M
$20 / 1M

Best for: Hard, multi-step work you would pay extra for: long coding sessions, research write-ups, and jobs that run for a while.

Paid Claude plans (Pro and Max). Official consumer guide: Free includes Haiku and Sonnet; Pro and Max add Opus and Fable.

Source: Opus 5.5 overview

Claude Sonnet 5.5

Current
Context
1M
Max output
128K
API input / 1M
$2 / 1M
API output / 1M
$10 / 1M

Best for: Day-to-day writing, coding, and analysis — the version most beginners should start with.

Free and paid Claude plans. Official consumer guide lists Sonnet on Free, Pro, and Max.

Source: Sonnet 5.5 overview

Claude Haiku 5.5

Current
Context
1M
Max output
128K
API input / 1M
$0.1 / 1M
API output / 1M
$0.5 / 1M

Best for: Quick answers, sorting, extraction, and anything you run hundreds of times.

Free and paid Claude plans. Official consumer guide lists Haiku on Free, Pro, and Max.

Source: Haiku 5.5 overview

Claude Fable 5.1

Current
Context
1M
Max output
128K
API input / 1M
$10 / 1M
API output / 1M
$50 / 1M

Best for: The rare job where Opus is not enough — long autonomous work you are willing to pay more for.

Paid Claude plans. Official consumer guide: Pro and Max add Fable; Anthropic’s pricing table also lists usage-credit / weekly-limit rules on Max.

Source: Models overview

Claude Opus 5.5

Hard, multi-step work you would pay extra for: long coding sessions, research write-ups, and jobs that run for a while.

  • Anthropic’s current default pick for long coding and knowledge-work jobs on the API.
  • 1 million token API context window — enough for a large project in one go.
  • Vision plus tools: it can read images and call tools, not just chat.

Caveats

  • On claude.ai the listed consumer context window is 200k tokens, not the 1M API window.
  • Uses more of a paid plan’s usage limit than Sonnet or Haiku.
  • Adaptive thinking cannot be turned off on this model.

Claude Sonnet 5.5

Day-to-day writing, coding, and analysis — the version most beginners should start with.

  • Anthropic’s stated best mix of speed and quality.
  • Same 1M API context and 128k max output as Opus, at half the API price.
  • Available on the free claude.ai plan, so you can try it without paying.

Caveats

  • On claude.ai the listed consumer context window is 200k tokens.
  • For the hardest all-day jobs, Anthropic still points people to Opus or Fable.

Claude Haiku 5.5

Quick answers, sorting, extraction, and anything you run hundreds of times.

  • Cheapest current Claude on the API (from $0.10 / $0.50 per 1M tokens).
  • Fastest in Anthropic’s current lineup.
  • Same 1M API context window as the larger models.

Caveats

  • API price jumps when the prompt is over 100,000 tokens.
  • Not the model Anthropic recommends for hard, long-running work.

Prompts up to 100k tokens. Prompts over 100k are $0.50 / $2.50 per 1M input/output.

Claude Fable 5.1

The rare job where Opus is not enough — long autonomous work you are willing to pay more for.

  • Anthropic’s current top model for the longest, hardest agent jobs.
  • Same 1M API context and 128k max output as the rest of the current lineup.

Caveats

  • Most expensive current Claude ($10 / $50 per 1M tokens on the API).
  • Slower than Opus, Sonnet, and Haiku in Anthropic’s own comparison table.
  • Thinking is always on.

Catalog pages