Model Selection

Claude Code lets you pick the model per session. As of October 2026 the current lineup is the 5.5 family plus Fable at the top:

Model	ID	In Claude Code since	Default effort in Claude Code
Fable 5.1	claude-fable-5-1	2.1.257 (2026-09-01)	high
Opus 5.5	claude-opus-5-5	2.1.280 (2026-09-22)	medium
Sonnet 5.5	claude-sonnet-5-5	2.1.284 (2026-09-28)	medium
Haiku 5.5	claude-haiku-5-5	2.1.293 (2026-10-07)	medium

All four have a 1M-token context window and 128K max output. Each 5.5 model became the default for its alias (opus, sonnet, haiku) on the Anthropic API in the release that added it. Fable is never the account default; you pick it in /model, and depending on your plan it can bill usage credits (Claude Code asks first). Prices are on Models & Cost.

The default is now Opus everywhere. Since 2.1.280 the default model is Opus 5.5 on Pro, Max, Team, Enterprise and the Anthropic API, and on Bedrock, Google Cloud and Claude Platform on AWS. That release moved Pro and Team Standard from Sonnet to Opus, matching Max, Team Premium and Enterprise. Microsoft Foundry still defaults to Sonnet 4.5. An admin-set organization default, or a model recorded on your account, takes precedence.

The practical rule still holds: match the model to the hardest step, not the whole task. Opus for architecture, gnarly debugging and multi-file reasoning; Sonnet for routine implementation; Haiku for tight, well-specified edits and high-volume fan-out; Fable when Opus at higher effort still falls short.

Effort, per model

Effort is set per model and saved per model. The three 5.5 models start at medium, one level below where Opus 5 and Sonnet 5 started. Anthropic's Claude Code docs say Opus 5.5 at medium matches or beats Opus 5 at high on their coding and knowledge-work evaluations, and that it thinks more per turn at a given level. So when you move up from a 5 model, start at medium rather than carrying over your old level.

• Old saved levels do not carry over. A level saved before /effort became per-model (including a top-level effortLevel in your user settings) no longer applies to Opus 5.5 and other new models; they start at their default until you pick one (2.1.280).
• Every current model takes low to max. /effort sets it; Enter saves it as your default for that model, and max applies to the current session only.
• Ultracode is its own toggle now. Since 2.1.284 it sits beside the level in /effort (Tab, or /effort ultracode [on|off]) and no longer forces xhigh. See Workflows.

How you use it day-to-day

• Switch with /model. It changes the current session only. Press d in the picker to set the default for new sessions. Change tiers as the work changes instead of committing up front.
• Fast mode (/fast) keeps Opus but speeds up output, up to 2.5x faster per the docs. It does not downgrade to a smaller model. It works on Opus 5.5 (the default since 2.1.280), Opus 5 and Opus 4.8, costs twice the standard Opus 5.5 rate, and on Pro, Max, Team and Enterprise it bills usage credits rather than your plan's included usage.
• Set fallback models. The fallbackModel setting (and --fallback-model) takes up to three models tried in order when your primary is overloaded or unavailable, so a busy Opus doesn't stall the session.
• Delegate cheap, decide expensive. When you fan out subagents for independent grunt work, run them on Sonnet or Haiku, and at lower effort (the Agent tool takes an effort parameter since 2.1.292), and keep the orchestrating session on Opus.

Gotchas

• Bigger isn't always better. Opus on trivial mechanical edits mostly burns tokens and time. Right-size to the step.
• Fast mode costs more for the whole conversation. The first time you turn it on in a conversation, you pay the fast uncached input price for the entire context so far. Turn it on early or not at all.
• Admins may gate new releases. With availableModelsMatch: "exact" or deniedModels in managed settings, a new model stays hidden until your admin lists it. See Permissions & Settings.
• Versions move fast. Before 5.5 the lineup was Opus 5, Sonnet 5 and Fable 5 (Fable 5.1 took over on September 1, and Opus 4.8 came before Opus 5); all remain available as legacy models. Treat the table above as an October 2026 snapshot and re-check the models overview and changelog rather than hard-coding a version.