The Claude model family

Claude isn't one model; it's a family, organized into tiers that trade off speed, cost, and raw capability. Picking the right one is a real skill, and it directly affects your results and your usage budget.

The tiers

Think of the tiers as different "brains" for different jobs:

Tier	Personality	Use it when…
Haiku	Fast & economical	Speed and volume matter more than depth
Sonnet	Balanced all-rounder	Everyday work, the sensible default
Opus	Deep & rigorous	Hard reasoning, important writing, complex analysis
Mythos-class (e.g. Fable)	Frontier intelligence	The hardest, long-horizon, high-stakes work

Mental model

• Haiku = a sharp intern who replies instantly.
• Sonnet = a capable colleague who handles most things well.
• Opus = a senior expert you bring in for the tough stuff.
• Mythos-class = the specialist you call for the truly hard problem (at a premium).

The lineup as of mid-2026

Model names and versions move quickly. As of mid-2026, the notable generally-available models are:

• Claude Haiku 4.5 (claude-haiku-4-5: fastest and most economical, with surprisingly strong reasoning. Great for real-time and high-volume tasks.
• Claude Sonnet 4.6 (claude-sonnet-4-6: the best balance of speed and intelligence for daily coding, analysis, and tool use.
• Claude Opus 4.8 (claude-opus-4-8: frontier-class flagship for complex coding, agents, and knowledge work. The go-to when quality matters.
• Claude Fable 5 (claude-fable-5: a Mythos-class model, a tier above Opus, built for the hardest long-horizon agentic work. Most capable, roughly 2× the cost of Opus.

All of these share a very large context window (around 1M tokens) and support advanced reasoning ("thinking").

This list will age

New versions ship regularly. Use this page to understand the tiers (Haiku/Sonnet/Opus/Mythos), then check the official model docs for the exact current versions.

How the tiers relate

            capability  ───────────────────────────►
 cost ▲     Haiku  ──►  Sonnet  ──►  Opus  ──►  Mythos (Fable)
 speed ▼    fastest                              slowest / priciest

Higher tiers are smarter but slower and more expensive per message. The art is using the lightest model that reliably does the job, and escalating only when you hit its limits.

Why this matters for you

• Cost & limits: Heavier models burn your usage budget faster (see Plans.
• Latency: Lighter models feel snappier, better for back-and-forth drafting.
• Quality ceiling: For genuinely hard reasoning, a top model can succeed where a lighter one quietly gets it wrong.

Next: a practical guide to choosing the right model for any given task.