claude-models

<a href="/en/glossary/core-concepts/anthropic/">Anthropic</a>'s flagship model officially released June 9, 2026 — the strongest broadly-available Claude model. 1M token <a href="https://aiagent-bible.com/en/glossary/agent-fundamentals/agent-context-window/" target="_blank" rel="noopener">Context Window</a>, 128K max output, <a href="/en/glossary/core-concepts/adaptive-thinking/">Adaptive Thinking</a> always enabled, priced at $10/M input and $50/M output. Designed for 'the most demanding reasoning and long-horizon agentic work' — the capability ceiling above Opus 4.8. Note: Fable 5 does not support <a href="/en/glossary/core-concepts/extended-thinking/">Extended Thinking</a>, only <a href="https://claudeskill-me.com/en/glossary/prompt-engineering/adaptive-thinking/" target="_blank" rel="noopener">Adaptive Thinking</a>.

The fastest and least expensive model in the Claude lineup. Built for high-volume simple tasks — translation, classification, format conversion, quick Q&A. Use it when you need to handle hundreds of requests per second, or when the task is simple enough that you don't need "smart."

<a href="/en/glossary/core-concepts/anthropic/">Anthropic</a> designs Claude in three tiers: Haiku (fastest and cheapest), Sonnet (best balance of capability and cost), and Opus (most capable but most expensive). Each suits different task types — it's not 'more expensive is always better.' Sonnet handles 90% of everyday tasks; Haiku suits high-frequency simple tasks; only a small number of tasks requiring extreme deep reasoning need Opus.

<a href="/en/glossary/core-concepts/anthropic/">Anthropic</a>'s flagship model at the top of the three-tier model gradient, known for the strongest complex reasoning, sustained long logic chains, and high-difficulty multi-step tasks. Costs 5× Sonnet and 20× Haiku; suited for scenarios "only Opus can handle well" — maintaining 5+ step reasoning chains, difficult multi-file architecture design, or complex analysis requiring rigorous argumentation. Current version: Claude Opus 4.

<a href="/en/glossary/core-concepts/anthropic/">Anthropic</a>'s flagship mid-tier model, positioned as the "best default choice for 90% of scenarios" — capability approaching Opus, better speed and cost than Opus. Claude 4 era's Sonnet 4.5 surpasses the previous generation Opus on complex reasoning, code, and long-form writing. Currently the highest API call volume Claude model.