Claude Models: Latest Lineup Compared (October 2026)

The latest Claude models are Claude Opus 5.5 and Claude Sonnet 5.5, alongside Haiku 4.5 and Fable 5.1. Opus 5.5 ($4 / $20 per million tokens) is Anthropic's recommended default, Sonnet 5.5 is the faster mid-price option, Haiku 4.5 is the fastest and cheapest, and Fable 5.1 is the most capable. All four accept image input.

What is the latest Claude model?

Claude Opus 5.5 and Claude Sonnet 5.5 are the newest releases. They replace Opus 5 and Sonnet 5 as the current Opus and Sonnet. Anthropic's own recommendation is to start with Opus 5.5 for most workloads and move up to Claude Fable 5.1 only when Opus 5.5 at higher effort still falls short.

The full current lineup is four models:

All four accept text and image input, produce text output, and support tool use and multilingual prompts. Relative speed runs Haiku 4.5 (fastest), Sonnet 5.5, Opus 5.5, then Fable 5.1 (slowest).

What changed in Opus 5.5 and Sonnet 5.5?

Opus 5.5 is cheaper than the model it replaces. Opus 5 was $5 / $25 per million tokens; Opus 5.5 is $4 / $20, with the same 1M context window and 128K max output. Two behavior changes matter if you're migrating API code: thinking can no longer be turned off (the effort setting is the only control), and the default effort dropped from high to medium. Set effort explicitly if your results feel shallower after switching.

Sonnet 5.5 keeps Sonnet 5's price at $2 / $10 and moves the knowledge cutoff forward to June 2026, from January 2026 on Sonnet 5. Effort still defaults to high, but the levels are recalibrated, so re-test the setting you used on Sonnet 5.

Neither Opus 5 nor Sonnet 5 is going away soon. Both are now listed as legacy models and stay available, but new projects should use the 5.5 versions.

Which Claude model should I use?

Start with Claude Opus 5.5 unless you have a specific reason not to. It's Anthropic's recommended default, coding included, and it's now cheaper than Opus 5 was.

Move to Claude Fable 5.1 when Opus 5.5 at higher effort genuinely isn't enough: a task that needs sustained reasoning across many steps, or an eval suite where Opus 5.5 plateaus below your bar. Fable 5.1 costs 2.5x as much per token as Opus 5.5, so make it a measured decision.

Pick Claude Sonnet 5.5 for production traffic where you want near-Opus quality at half the price, or where latency matters more than the last few points of capability.

Pick Claude Haiku 4.5 for anything high-volume and latency-sensitive: classification, routing, autocomplete-style agent loops, first-pass triage before escalating to a bigger model. It's the cheapest and fastest model in the lineup, with a smaller 200K context window.

Which Claude model is best for images?

Every current Claude model can read images, including screenshots, photos, charts, diagrams, and scanned PDFs. No Claude model generates images. If you need image generation, Claude isn't the tool; if you need image understanding, any of the four works.

How to choose for image work:

• Screenshots, UI review, charts, and documents: Claude Opus 5.5, the recommended default. It handles the common jobs well, like turning a UI screenshot into code or pulling numbers out of a chart.
• Hardest visual reasoning: Claude Fable 5.1, for dense technical diagrams or multi-image analysis where Opus 5.5 misses details.
• High-volume image tagging or moderation: Claude Haiku 4.5. At $1 / $5 per million tokens, it's the economical choice when you're processing thousands of images and each judgment is simple.
• Production apps with image input at scale: Claude Sonnet 5.5, which balances quality and cost.

What are the Claude model API names?

Since the Claude 4.6 generation, model IDs have no snapshot date: claude-{name}-{major}-{minor}, such as claude-opus-5-5. The same string works on the Claude API, Claude Platform on AWS, Google Cloud, and Microsoft Foundry. Amazon Bedrock adds an anthropic. prefix.

Claude Haiku 4.5 is the exception. It predates the dateless format, so its full API name is claude-haiku-4-5-20251001. The alias claude-haiku-4-5 resolves to that snapshot, and Google Cloud uses an @ separator: claude-haiku-4-5@20251001.

How much do Claude models cost?

Base Claude API pricing per million tokens:

Batch API requests are 50% off on every model. Prompt caching matters more than the headline price for any workload with a large, stable system prompt or repeated context. Run your actual token mix through the plan calculator before picking a model on price alone.

Which Claude models are legacy or being retired?

These older models are still available on the Claude API, but none of them is the right pick for a new project: Claude Fable 5, Claude Opus 5, Claude Opus 4.8, Claude Opus 4.7, Claude Opus 4.6, Claude Opus 4.5, Claude Sonnet 5, and Claude Sonnet 4.6.

Claude Sonnet 4.5 is deprecated and retires on November 30, 2026. Anthropic's recommended replacement is claude-sonnet-5-5.

Retirement commitments for the current four, on Anthropic-operated platforms (Bedrock and Google Cloud set their own dates):

• Claude Haiku 4.5: not sooner than October 15, 2026
• Claude Fable 5.1: not sooner than September 1, 2027
• Claude Opus 5.5: not sooner than September 22, 2027
• Claude Sonnet 5.5: not sooner than September 28, 2027

Haiku 4.5's date is the one to watch. No deprecation has been announced yet, and Anthropic gives at least 60 days' notice before retiring a publicly released model. Still, a high-volume Haiku workload should have a migration plan ready.

The mistake almost everyone makes

Treating a model ID as if it pointed to "whatever is newest." It doesn't. Every Claude model ID, including dateless ones like claude-opus-5-5, is a pinned snapshot. The weights behind it never change. When Anthropic ships an improved model, it gets a new ID, and your integration keeps running exactly what it was tested against until you change the string yourself.

So an app built on Opus 5 or Sonnet 5 stays there after the 5.5 releases. If you want the newest model, you have to switch to it: update the model string in your API code, or use /model in Claude Code (see the Claude Code commands reference) to switch the active model mid-session.

FAQ

Keep reading