Claude’s `ultrathink` Deprecated: What’s New and a Hidden Trick for 64K Models

If you’ve been using Anthropic's Claude, you might remember the ultrathink keyword that unlocked its maximum reasoning power. Well, it’s no longer needed. But there’s a new default and even a hidden trick to get twice the thinking budget on 64K output models. Let’s dive in.

The Old Way: The ultrathink Keyword

For months, adding ultrathink to your prompt would give you 31,999 thinking tokens. Power users often combined it with "Opus + Ultrathink + Plan Mode" for optimal results. Under the hood, Claude’s code detected the keyword and set the thinking budget:

// Simplified (other keywords like "megathink" and "think" existed too)
const thinkingBudget = prompt.includes("ultrathink") ? 31999 : 0;

// Passed to the Anthropic API
await client.messages.create({
  model: "claude-sonnet-4-...",
  messages: [...],
  thinking: thinkingBudget > 0 ? {
    type: "enabled",
    budget_tokens: thinkingBudget  // This is what mattered
  } : undefined
});

What Changed

On January 16, 2026, Anthropic deprecated ultrathink:

“Closing as ultrathink is now deprecated and thinking mode is enabled by default.” , Sarah Deaton, Anthropic

If you type "ultrathink" now, you’ll see a deprecation message.

The New Default: Automatic Extended Thinking

Extended thinking is now automatically enabled for supported models with a default budget of 31,999 tokens. Here are the supported models:

• Opus 4.5 ✓
• Sonnet 4.5 ✓
• Sonnet 4 ✓
• Haiku 4.5 ✓
• Opus 4 ✓
• Claude 3.x ✗ (not supported)

The new API call flow looks like this:

// 1. Determine thinking budget
let budgetTokens = 31999;  // Default: max

if (process.env.MAX_THINKING_TOKENS) {
  budgetTokens = parseInt(process.env.MAX_THINKING_TOKENS);
}

// 2. Auto-enabled for supported models
const thinkingEnabled = isSupportedModel(model);  // Opus 4.5, Sonnet 4/4.5, Haiku 4.5, Opus 4

// 3. Passed to Anthropic API on every request
await client.messages.create({
  model: "claude-sonnet-4-...",
  messages: [...],
  thinking: thinkingEnabled ? {
    type: "enabled",
    budget_tokens: budgetTokens  // Default is 31,999
  } : undefined
});

The Hidden Unlock: 63,999 Tokens

Here’s a lesser-known fact: the 31,999 default exists for backward compatibility with Opus 4 (which has a 32K output limit). However, 64K output models can support much more:

| Model | Max Output | Max Thinking Budget | | --- | --- | --- | | Opus 4.5 | 64,000 | 63,999 | | Sonnet 4.5 | 64,000 | 63,999 | | Sonnet 4 | 64,000 | 63,999 | | Haiku 4.5 | 64,000 | 63,999 | | Opus 4 | 32,000 | 31,999 |

The thinking budget is capped at max_tokens - 1 because budget_tokens must be strictly less than max_tokens, leaving at least one token for the output.

Wrapping Up

You no longer need to use magic keywords like `ultrath