← All guidesThe models
- Claude Haiku 4.5 - the fast one. Snappy answers for everyday back-and-forth at the lowest cost.
- Claude Haiku 5.5 - the newest fast one. Quick replies at the lowest price of any Claude model, with a 1M-token context window.
- Claude Sonnet 4.6 - the all-rounder. Strong reasoning at sensible speed; the sweet spot for daily work.
- Claude Sonnet 5.5 - the new all-rounder. Fast, strong reasoning for less than Sonnet 4.6.
- Claude Opus 4.6 - the heavyweight. Deepest reasoning for hard problems, long documents and gnarly plans.
- Claude Opus 5.5 - the flagship Opus. Anthropic's recommended starting point for serious work.
- Claude Fable 5.1 - the top tier. Anthropic's most capable model for the hardest reasoning, and the model Genius mode pins every Claude agent to.
- Claude Fable 5.2, Claude Opus 6 and Claude Haiku 6 - expected soon; they join the roster the day Anthropic releases them.
What only Claude does in Magnus
Three capabilities are Claude-specific today: live web search with citations (the model searches mid-answer when a question depends on current information), native PDF reading (attach a PDF and Claude reads pages as pages - tables, layout and all), and the full integration toolset, where Magnus reads your Gmail, books calendar slots, posts to Slack and more as part of a single answer.
Getting it
Claude is included in your plan - just pick it (or add your own Anthropic key in Settings). You can also run different agents on different Claude models: Haiku for the quick-fire persona, Opus for the strategist.
- If you connect integrations, run your main agent on Claude - other providers answer chat fine but skip the tools.