Claude models in 2026: Opus, Sonnet, Haiku, or Fable

Claude models: the August 2026 lineup

Claude Opus 5 is the coding flagship at $5 input and $25 output per million on the API. Fable 5 sits above it for runs larger than one sitting. How Fable is billed on a Claude plan is in the table below, from Anthropic's Fable plan article. The Fable API table is also on Claude Code pricing. Session and weekly windows shared across models live on Claude Code limits. How those Claude models sit against GPT-5.6 and Gemini is on best coding models.

• Output is priced five times input on every current model, so the output column is where the difference actually bites.
• Cache reads bill at 10 percent of the base input rate, which is why a long cached session is far cheaper than the input column alone suggests.
• The Batch API halves both input and output, but it is asynchronous and irrelevant to interactive coding.
• Haiku 4.5 does not support effort levels. Fable 5, Opus 5, and Sonnet 5 all do, and default to high.

Coding-agent cards with harness, DeepSWE cost, and when not to use it: Opus 5.5, Sonnet 5.5, Opus 5, and Fable 5.

What your account defaults to

The default alias is not one model. Claude Code documentation, opened 19 August 2026, maps it by account type. Fable 5 is never the default on any of these rows. Sessions use Fable only after you choose it with /model fable, a model setting, or the best alias where Fable is available.

Fable 5 on your Claude plan

A promotion that included Fable 5 in Pro weekly limits ended 19 July 2026 at 11:59:59 PM PT. Starting 20 July 2026, access depends on the plan. This table is from Anthropic's Fable plan article and claude.com/pricing, opened 19 August 2026. Claude Code needs v2.1.170 or later.

Fable 5 is not available under zero data retention. The /model picker either omits it or shows it disabled. Source: Claude Code model configuration, opened 19 August 2026.

When Fable 5 bills usage credits, the /model picker shows "Requires usage credits" on the Fable 5 row. Interactive sessions get a consent prompt before that request bills. Members of Enterprise plans with organization billing do not see the prompt. After you choose to continue on Fable 5 using usage credits, the prompt does not return. In non-interactive mode with the -p flag, and through the Agent SDK, Claude Code never shows the consent prompt and bills without asking. Source: Claude Code model configuration, opened 19 August 2026.

Older Claude models still available

Aliases, and why yours may not mean what you think

Claude Code takes either a full model name or an alias. Aliases track the recommended version and change over time, which is convenient until you are on a provider where they resolve somewhere else.

1M context by plan

Claude Code's extended-context page, opened 19 August 2026, splits 1M access by plan and by model. On the Anthropic API, Fable 5, Sonnet 5, and Opus 4.7 and later always run with the 1M window. Sonnet 5 is native 1M on that API: no credit gate, no 200K variant, and no [1m] suffix to select.

• Max, Team, and Enterprise auto-upgrade Opus to 1M with no extra configuration.
• Pro needs usage credits for Opus 1M.
• Sonnet 5 on the Anthropic API is native 1M on every plan. There is no credit gate.
• opusplan[1m] forces 1M on both the plan-mode Opus phase and the execute-mode Sonnet phase. Use it when you are not on an auto-upgrade tier and want both phases at 1M.
• The 1M window uses standard model pricing. There is no extra rate past 200K.

Effort levels

Effort controls adaptive reasoning: how often and how deeply the model thinks on each step. Available levels come from Claude Code model configuration, opened 19 August 2026. Models not in this table do not support effort, except Haiku 4.5, which is listed so the none row is explicit. Default is high except Opus 4.7, which defaults to xhigh.

Thinking cannot be turned off on Fable 5. The session toggle, alwaysThinkingEnabled, and MAX_THINKING_TOKENS=0 have no effect there. Source: Claude Code model configuration, opened 19 August 2026.

Automatic model fallback

Fable 5 and Opus 5 run with safety classifiers for cybersecurity and biology content. When a classifier flags a request and the flagged category has a fallback model, Claude Code re-runs the request on that model and shows a notice in the transcript. After a fallback, the session continues on the fallback model. To return to your original model, run /model. Category-based fallback requires Claude Code v2.1.219 or later. Before v2.1.219, every flagged Fable 5 request re-ran on your provider's default Opus model, and Opus 5 was not a fallback source. Source: Claude Code model configuration, opened 19 August 2026.

To decide what happens each time a request is flagged, rather than switching automatically, run /config and turn off Switch models when a message is flagged, or set switchModelsOnFlag to false in your settings file. A flagged request then pauses the session with two options: switch to the fallback model, or edit the prompt and retry on the current model. Source: Claude Code model configuration, opened 19 August 2026.

• A biology flag on Opus 5 has no fallback. Claude Code does not show the prompt and the request ends with the refusal.
• When the fallback target is blocked by availableModels, Claude Code does not show the prompt. The flagged request ends with the refusal.
• If both models flag the same request, edit the prompt and retry, or start a new session.
• On mobile Claude Code on the web, editing and retrying is not supported. Switch models, or continue from a desktop browser or the desktop app.
• In non-interactive mode and SDK integrations that cannot show the prompt, a flagged request ends the turn with a refusal.

Fallback model chains

This is a different mechanism from the classifier table above. When the primary model is overloaded, unavailable, or returns another non-retryable server error, Claude Code can switch to a fallback model instead of failing the request. Authentication, billing, rate-limit, request-size, and transport errors never trigger a switch. Those follow their normal retry and error handling. Source: Claude Code model configuration, opened 19 August 2026.

Configure one or more fallback models and Claude Code tries them in order, showing a notice when it switches. The switch lasts for the current turn only, so your next message tries the primary model first again. Claude Code caps chains at three models after duplicate removal and ignores extra entries. Set a chain for one session with --fallback-model (comma-separated list). To persist a chain, set fallbackModel in settings as an array. The flag takes precedence over the setting.

claude --fallback-model sonnet,haiku

{
  "fallbackModel": ["claude-sonnet-5", "claude-haiku-4-5"]
}

The chain also covers compaction, but Claude Code will not fall back to a model with a smaller context window than the primary. If every fallback is smaller, compaction shows the original error and you can retry. Entries outside availableModels are dropped when the chain is read.

Setting and switching it

# 1. in session, effective from the next turn
/model sonnet
/model claude-opus-5-5
/model               # opens the picker; Enter saves as default, s is session-only

# 2. at launch, this session only
claude --model opus

# 3. environment, this session only
ANTHROPIC_MODEL=claude-haiku-4-5 claude

# 4. settings file, persistent
#   { "model": "claude-sonnet-5" }

• As of v2.1.153, /model in the picker: Enter saves the default for new sessions. s is this session only. A model set with /model in non-interactive -p applies to the current session only and is not saved.
• Picker prices appear only when Claude Code talks to the Anthropic API, directly or through a proxying LLM gateway. On Amazon Bedrock and the Claude apps gateway, the rows show no price. The price is a display label only.
• The picker asks for confirmation when the conversation already has output, because the next response re-reads the full history without cached context. That confirmation is a real cost warning, not a formality.
• Resumed sessions keep the model they were using when the transcript was saved, so another terminal switching models cannot change yours on resume.
• To run different models in different terminals at once, launch each with its own --model flag rather than switching with /model.
• A subagent can pin its own model in frontmatter, and a skill can override the model for the turn it runs in.

A routing table

What the gap costs on one real task

List rates per million tokens are hard to reason about, because an agent session is mostly cache reads. Here is the same task priced four ways, so the ratio is concrete rather than notional.

The task: 300,000 input tokens across the session, of which 90 percent arrive as cache reads, plus 20,000 output tokens. That is a realistic shape for a bounded feature or a substantial debugging run.

On a subscription the same ratios apply to your usage window rather than to a dollar figure: a session or weekly allowance is consumed roughly in proportion to token cost, and the windows are shared across models. Switching to Sonnet with /model after hitting an Opus-specific limit keeps you working; it does not restore a session or weekly window, because those are not per model.

Escalate on evidence, not on anxiety

Anthropic publishes two pieces of guidance that read differently, and both are correct in context. The platform models page says to start with Opus 5 for complex agentic coding and enterprise work. The Claude Code cost guidance says Sonnet handles most coding tasks and to reserve Opus for architectural decisions and multi-step reasoning. The first is answering "which model is capable of this"; the second is answering "which model should run your Tuesday".

The efficient pattern resolves both. It is not choosing a model per project, it is starting on the cheaper capable model your account actually has and escalating the moment a task has shown you it is hard.

1. If your account defaults to Sonnet 5, stay there for routine work. Pro and Team standard start on Sonnet. Max and API start on Opus 5, so drop to Sonnet with /model sonnet when the task does not need Opus.
2. Escalate to Opus 5 when Sonnet has produced a wrong answer twice, when the task is architectural, or when you cannot describe the fix yourself.
3. Drop to Haiku 4.5 the moment a task turns mechanical, which frequently happens partway through: Opus decides the approach, Haiku applies it across thirty files.
4. Escalate for review even when the work was done cheaply. A fresh Opus read of a finished diff in a subagent is one of the best uses of the expensive model there is.
5. Or let opusplan do it. It runs Opus during plan mode and Sonnet for execution, which is the same discipline with no manual switching.

Model choice also interacts with reasoning effort, and the two multiply. High effort on Opus 5 is many times the cost of the default on Sonnet 5 for work the second combination often completes the same way. Decide the model first, then the effort, and change one dial at a time.

Questions people ask

Keep reading

Continue through Claude Code reference with Claude Code best practices that survive contact with real work.